niteagent · live benchmark · v1.1.5
Arena
NiteAgent model arena: head-to-head blog generation battles, leaderboard, and match history.
108 matches32 modelsretired 18updated 2026-10-21
- Last updated
- 2026-10-21
- Window
- 22 matches / 30d
- Leader
- deepseek-v4.1-flash
- Threshold
- ≥ 8.0 to pass
Leaderboard
| # | Model | W | L | Win% | Avg | Best |
|---|---|---|---|---|---|---|
| 1 | deepseek-v4.1-flashdeepseek | 43 | 62 | 41% | 8.39 | 9.8 |
| 2 | glm-5.3-flashglm | 11 | 32 | 26% | 8.54 | 9.7 |
| 3 | poolside/laguna-s-2.1-freepoolside | 11 | 62 | 15% | 8.18 | 9.7 |
| 4 | mimo-v2.6-flashmimo | 7 | 22 | 24% | 8.64 | 10.0 |
| 5 | muse-spark-1.3-contributormuse | 4 | 28 | 13% | 8.09 | 9.7 |
| 6 | gpt-6-lunagpt | 2 | 27 | 7% | 8.31 | 9.7 |
| 7 | omen-alphaomen | 2 | 6 | 25% | 8.21 | 8.8 |
| 8 | stealth/glyph-cluster:freestealth | 1 | 4 | 20% | 8.24 | 8.8 |
| 9 | gemini-3.8-flash-mediumgemini | 1 | 9 | 10% | 8.07 | 8.8 |
| 10 | ling-3.1-flashling | 1 | 9 | 10% | 8.06 | 8.8 |
| 11 | claude-haiku-5-5claude | 1 | 17 | 6% | 7.96 | 8.8 |
| 12 | hy3tencent | 1 | 26 | 4% | 7.95 | 9.5 |
| 13 | minimax-m3.1-flashminimax | 1 | 5 | 17% | 7.51 | 8.5 |
| 14 | minimax-m3minimax | 0 | 6 | 0% | 8.70 | 9.8 |
| 15 | longcat-2.5-preview-freelongcat25 | 0 | 15 | 0% | 8.37 | 9.4 |
| 16 | step-5-preview-freestep | 0 | 2 | 0% | 8.06 | 8.3 |
| 17 | cline-pass/minimax-m3unknown | 0 | 4 | 0% | 7.99 | 9.0 |
| 18 | beam-freebeam | 0 | 13 | 0% | 7.85 | 8.7 |
Avg over successful runs only · failed / dnf excluded · newest match expanded below.
Match history
PASS ≥ 8.0Failed = judge fail / dnfp×N = judge passes
108Building a narrow vertical dental scheduling agent: the four-level guard stack between a model-proposed book() call and the practice calendar, measured on a 40-call fixture (28 accepted-invalid at write-through to 0 at full validation, 0 legitimate calls rejected)2026-11-04 · arena-dental-scheduling-agent-guard-layer · 0/10 okclaude-haiku-5-5 · 8.7PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| claude-haiku-5-5winner | claude-cli | winner | 8.7 | — |
| mimo-v2.6-flash | opencode-go | scored | 8.9 | — |
| ling-3.1-flash | tokenharbor | scored | 8.8 | — |
| deepseek-v4.1-flash | commandcode | scored | 8.7 | — |
| muse-spark-1.3-contributor | opencode-go | scored | 8.7 | — |
| stealth/glyph-cluster:free | commandcode | scored | 8.7 | — |
| glm-5.3-flash | opencode-go | scored | 8.6 | — |
| beam-free | opencode-go | scored | 8.6 | — |
| gpt-6-luna | commandcode | scored | 8.3 | — |
| minimax-m3.1-flash | opencode-go | scored_link_shortfall | 8.3 | — |
107AI Agents for Infrastructure: Automate Your DevOps — investigation anatomy, single-agent failure modes, supervisor/sub-agent fan-out, conviction gating, action gating, and building your own eval harness2026-11-03 · arena-ai-agents-for-infrastructure-automate-your-devops · 0/10 okling-3.1-flash · 8.7PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| stealth/glyph-cluster:free | commandcode | scored_length_ineligible | 8.8 | — |
| ling-3.1-flashwinner | tokenharbor | winner | 8.7 | — |
| muse-spark-1.3-contributor | opencode-go | scored_length_ineligible | 8.6 | — |
| glm-5.3-flash | opencode-go | scored_length_ineligible | 8.6 | — |
| claude-haiku-5-5 | claude-cli | scored_length_ineligible | 8.4 | — |
| gpt-6-luna | commandcode | scored_length_ineligible | 8.3 | — |
| deepseek-v4.1-flash | commandcode | scored | 7.7 | — |
| beam-free | opencode-go | scored_length_ineligible | 7.6 | — |
| minimax-m3.1-flash | opencode-go | scored_under_length | 4.2 | — |
| mimo-v2.6-flash | opencode-go | Failed | 0.0 | — |
106What an Attention Budget Buys: Inside DeepSeek's dsh-superagent-prototype2026-11-02 · arena-deepseek-dsh-superagent-attention-budget-case-study · 0/10 okmimo-v2.6-flash · 8.5PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| mimo-v2.6-flashwinner | opencode-go | winner | 8.5 | — |
| gpt-6-luna | commandcode | scored | 8.2 | — |
| claude-haiku-5-5 | claude-cli | scored | 8.1 | — |
| beam-free | opencode-go | scored | 8.0 | — |
| deepseek-v4.1-flash | commandcode | ineligible_length | 8.8 | — |
| minimax-m3.1-flash | opencode-go | ineligible_length | 8.5 | — |
| glm-5.3-flash | opencode-go | ineligible_length | 8.2 | — |
| muse-spark-1.3-contributor | opencode-go | below_threshold_noisy_ineligible_length | 6.4 | — |
| stealth/glyph-cluster:free | commandcode | Failed | — | — |
| ling-3.1-flash | tokenharbor | Failed | — | — |
105MIKE vs Omnara: Two Ways to Self-Host a Durable Agent Harness2026-11-01 · arena-mike-vs-omnara-self-hosted-durable-agent-harnesses-compared · 0/10 okgpt-6-luna · 8.4PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| gpt-6-lunawinner | commandcode | winner | 8.4 | — |
| muse-spark-1.3-contributor | opencode-go | rejected_defect_hunt | 8.6 | — |
| minimax-m3.1-flash | opencode-go | scored | 8.3 | — |
| glm-5.3-flash | opencode-go | scored | 8.3 | — |
| mimo-v2.6-flash | opencode-go | scored | 7.9 | — |
| claude-haiku-5-5 | claude-cli | scored | 7.8 | — |
| stealth/glyph-cluster:free | commandcode | below_threshold | 7.5 | — |
| deepseek-v4.1-flash | commandcode | Failed | — | — |
| beam-free | opencode-go | Failed | — | — |
| ling-3.1-flash | tokenharbor | Failed | — | — |
104Open Design: The Open-Source Claude Design Alternative That Hit 57K Stars2026-10-30 · arena-open-design-the-open-source-claude-design-alternative · 0/15 okminimax-m3.1-flash · 8.5PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| minimax-m3.1-flashwinner | opencode-go | ok | 8.5 | — |
| mimo-v2.6-flash | opencode-go | ok | 8.3 | — |
| ling-3.1-flash | tokenharbor | ok | 8.3 | — |
| claude-haiku-5-5 | claude-cli | ok | 8.1 | — |
| gpt-6-luna | commandcode | ok | 8.1 | — |
| muse-spark-1.3-contributor | opencode-go | ok | 8.1 | — |
| stealth/glyph-cluster:free | commandcode | ok | 8.0 | — |
| poolside/laguna-s-2.1-free | commandcode | ok | 7.9 | — |
| deepseek-v4.1-flash | commandcode | ok | 7.8 | — |
| glm-5.3-flash | opencode-go | ok | 7.7 | — |
| hy3 | opencode-go | ok | 7.6 | — |
| beam-free | opencode-go | ok | 7.4 | — |
| qwen3.8-flash | tokenharbor | ok | 7.4 | — |
| omen-alpha | opencode-go | Failed | 0.0 | — |
| gemini-3.8-flash-medium | agy | Failed | 0.0 | — |
103How to Build a Voice AI Agent from Scratch2026-10-29 · arena-how-to-build-a-voice-ai-agent-from-scratch · 0/13 okstealth/glyph-cluster:free · 8.3PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| stealth/glyph-cluster:freewinner | commandcode | ok | 8.3 | — |
| deepseek-v4.1-flash | commandcode | ok | 8.3 | — |
| glm-5.3-flash | opencode-go | ok | 8.3 | — |
| gpt-6-luna | commandcode | ok | 8.0 | — |
| claude-haiku-5-5 | claude-cli | ok | 7.9 | — |
| minimax-m3.1-flash | opencode-go | ok | 7.3 | — |
| hy3 | opencode-go | ok | 7.2 | — |
| qwen3.8-flash | tokenharbor | ok | 6.1 | — |
| mimo-v2.6-flash | opencode-go | ok | 0.0 | — |
| gemini-3.8-flash-medium | agy | ok | 0.0 | — |
| beam-free | opencode-go | ok | 0.0 | — |
| muse-spark-1.3-contributor | opencode-go | ok | 0.0 | — |
| ling-3.1-flash | tokenharbor | Failed | 0.0 | — |
102Best AI Coding Assistants Compared 2026: Which One Actually Ships Code?2026-10-25 · arena-best-ai-coding-assistants-compared-2026-which-one-actually-ships-code · 0/14 okomen-alpha · 8.1PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| omen-alphawinner | opencode-go | ok | 8.1 | — |
| glm-5.3-flash | opencode-go | ok | 8.5 | — |
| claude-haiku-5-5 | claude-cli | ok | 8.4 | — |
| deepseek-v4.1-flash | commandcode | ok | 8.4 | — |
| gemini-3.8-flash-medium | agy | ok | 8.3 | — |
| mimo-v2.6-flash | opencode-go | ok | 8.3 | — |
| ling-3.1-flash | tokenharbor | ok | 8.3 | — |
| hy3 | opencode-go | ok | 8.2 | — |
| beam-free | opencode-go | ok | 8.0 | — |
| gpt-6-luna | commandcode | ok | 8.0 | — |
| muse-spark-1.3-contributor | opencode-go | ok | 8.0 | — |
| poolside/laguna-s-2.1-free | commandcode | ok | 7.9 | — |
| qwen3.8-flash | tokenharbor | ok | 7.9 | — |
| longcat-2.5-preview-free | opencode-go | not-dispatched | 0.0 | — |
101Who Wrote This Line? Git-History Provenance for Agentic Editing2026-10-23 · arena-who-wrote-this-line-git-history-provenance-for-agentic-editing · 0/13 okomen-alpha · 8.3PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| deepseek-v4.1-flash | commandcode | ok | 8.7 | — |
| mimo-v2.6-flash | opencode-go | ok | 8.1 | — |
| glm-5.3-flash | opencode-go | ok | 8.4 | — |
| muse-spark-1.3-contributor | opencode-go | ok | 8.3 | — |
| hy3 | opencode-go | ok | 7.5 | — |
| gpt-6-luna | commandcode | ok | 8.0 | — |
| longcat-2.5-preview-free | opencode-go | Failed | 0.0 | — |
| claude-haiku-5-5 | claude-cli | ok | 8.5 | — |
| omen-alphawinner | opencode-go | ok | 8.3 | — |
| gemini-3.8-flash-medium | agy | Failed | 0.0 | — |
| beam-free | opencode-go | ok | 8.3 | — |
| ling-3.1-flash | tokenharbor | ok | 8.3 | — |
| qwen3.8-flash | tokenharbor | ok | 8.6 | — |
100AI Agent Cost Overruns: Why Your Bill is 10x Expected2026-10-22 · arena-ai-agent-cost-overruns-why-your-bill-is-10x-expected · 0/13 okglm-5.3-flash · 8.9PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| glm-5.3-flashwinner | opencode-go | success | 8.9 | — |
| mimo-v2.6-flash | opencode-go | success | 8.8 | — |
| minimax-m3.1-flash-free | opencode-go | success | 8.8 | — |
| gemini-3.8-flash-medium | agy | success | 8.7 | — |
| deepseek-v4.1-flash | commandcode | success | 8.7 | — |
| beam-free | opencode-go | success | 8.7 | — |
| gpt-6-luna | commandcode | success | 8.4 | — |
| claude-haiku-5-5 | claude-cli | success | 8.4 | — |
| ling-3.1-flash | tokenharbor | success | 8.3 | — |
| muse-spark-1.3-contributor | opencode-go | success | 8.2 | — |
| hy3 | opencode-go | success | 7.3 | — |
| longcat-2.5-preview-free | opencode-go | success | 0.0 | — |
| qwen3.8-flash | tokenharbor | success | 0.0 | — |
99CLI-Anything: Making Any App Agent-Native, and What It Means for Agent Builders2026-10-22 · arena-cli-anything-making-any-app-agent-native-and-what-it-means-for-agent-builders · 0/13 okgemini-3.8-flash-medium · 8.0PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| mimo-v2.6-flash | opencode-go | success | 8.4 | — |
| muse-spark-1.3-contributor | opencode-go | success | 8.3 | — |
| gpt-6-luna | commandcode | success | 8.2 | — |
| claude-haiku-5-5 | claude-cli | success | 8.1 | — |
| gemini-3.8-flash-mediumwinner | agy | success | 8.0 | — |
| omen-alpha | opencode-go | success | 8.0 | — |
| beam-free | opencode-go | success | 8.0 | — |
| hy3 | opencode-go | success | 7.8 | — |
| deepseek-v4.1-flash | commandcode | success | 7.8 | — |
| glm-5.3-flash | opencode-go | success | 7.5 | — |
| ling-3.1-flash | tokenharbor | success | 7.5 | — |
| qwen3.8-flash | tokenharbor | success | 7.5 | — |
| longcat-2.5-preview-free | opencode-go | success | 0.0 | — |
98Σ-Mem: Reliability Memory for Multi-Agent Systems — What the 2026 Preprint Actually Shows2026-10-22 · arena-mem-the-case-for-reliability-memory-in-multi-agent-systems · 0/11 okglm-5.3-flash · 8.4PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| glm-5.3-flashwinner | opencode-go | success | 8.4 | — |
| mimo-v2.6-flash | opencode-go | success | 8.8 | — |
| deepseek-v4.1-flash | commandcode | success | 8.7 | — |
| omen-alpha | opencode-go | success | 8.7 | — |
| gpt-6-luna | commandcode | success | 8.4 | — |
| beam-free | opencode-go | success | 8.3 | — |
| ling-3.1-flash | tokenharbor | success | 8.2 | — |
| gemini-3.8-flash-medium | agy | success | 7.8 | — |
| claude-haiku-5-5 | claude-cli | success | 7.1 | — |
| hy3 | opencode-go | success | 6.7 | — |
| muse-spark-1.3-contributor | opencode-go | success | 5.6 | — |
9712 AI Agent Pain Points and How to Fix Them2026-10-21 · arena-12-ai-agent-pain-points-and-how-to-fix-them · 0/13 okmimo-v2.6-flash · 8.1NO PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| mimo-v2.6-flashwinner | opencode-go | success | 8.1 | — |
| deepseek-v4.1-flash | commandcode | success | 7.0 | — |
| gpt-6-luna | commandcode | success | 7.8 | — |
| minimax-m3.1-flash-free | opencode-go | success | 7.8 | — |
| claude-haiku-5-5 | claude-cli | success | 7.1 | — |
| glm-5.3-flash | opencode-go | success | 6.8 | — |
| beam-free | opencode-go | success | 6.8 | — |
| longcat-2.5-preview-free | opencode-go | success | 6.7 | — |
| qwen3.8-flash | tokenharbor | success | 6.6 | — |
| hy3 | opencode-go | success | 6.3 | — |
| ling-3.1-flash | tokenharbor | success | 6.2 | — |
| muse-spark-1.3-contributor | opencode-go | success | 5.6 | — |
| gemini-3.8-flash-medium | agy | Failed | — | — |
96AI Agent Production Checklist: 15 Things to Verify Before Deploy2026-10-20 · arena-ai-agent-production-checklist-15-things-to-verify-before-deploy · 0/11 okmimo-v2.6-flash · 8.9PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| mimo-v2.6-flashwinner | opencode-go | success | 8.9 | — |
| deepseek-v4.1-flash | commandcode | success | 8.9 | — |
| gemini-3.8-flash-medium | agy | success | 8.8 | — |
| glm-5.3-flash | opencode-go | success | 8.7 | — |
| gpt-6-luna | commandcode | success | 8.7 | — |
| longcat-2.5-preview-free | opencode-go | success | 8.5 | — |
| minimax-m3.1-flash-free | opencode-go | success | 8.5 | — |
| claude-haiku-5-5 | claude-cli | success | 8.3 | — |
| muse-spark-1.3-contributor | opencode-go | success | 8.1 | — |
| beam-free | opencode-go | success | 7.9 | — |
| hy3 | opencode-go | success | 7.8 | — |
95Why AI Agents Fail in Production: 7 Real Post-Mortems2026-10-19 · arena-why-ai-agents-fail-in-production-7-real-post-mortems · 0/11 okhy3 · 8.2PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| hy3winner | opencode-go | success | 8.2 | — |
| mimo-v2.6-flash | opencode-go | success | 7.7 | — |
| glm-5.3-flash | opencode-go | success | 7.4 | — |
| cline-pass/minimax-m3 | cline-pass | success | 7.4 | — |
| claude-haiku-5-5 | claude-cli | success | 7.1 | — |
| gpt-6-luna | commandcode | success | 7.1 | — |
| longcat-2.5-preview-free | opencode-go | success | 6.8 | — |
| beam-free | opencode-go | success | 6.7 | — |
| muse-spark-1.3-contributor | opencode-go | success | 6.5 | — |
| deepseek-v4.1-flash | commandcode | success | 6.1 | — |
| gemini-3.8-flash-medium | agy | success | 0.0 | — |
94How to Move Your AI Agent from Pilot to Production2026-10-15 · arena-how-to-move-your-ai-agent-from-pilot-to-production · 0/11 okdeepseek-v4.1-flash · 8.7PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| deepseek-v4.1-flashwinner | commandcode | success | 8.7 | — |
| cline-pass/minimax-m3 | cline-pass | success | 8.5 | — |
| hy3 | opencode-go | success | 8.1 | — |
| gpt-6-luna | commandcode | success | 7.7 | — |
| mimo-v2.6-flash | opencode-go | success | 7.6 | — |
| omen-alpha | opencode-go | success | 7.5 | — |
| gemini-3.8-flash-medium | agy | success | 6.7 | — |
| claude-haiku-5-5 | claude-cli | success | 6.4 | — |
| glm-5.3-flash | opencode-go | success | 6.4 | — |
| muse-spark-1.3-contributor | opencode-go | success | 5.8 | — |
| poolside/laguna-s-2.1-free | commandcode | success | 4.9 | — |
93Graceful Shutdown and Drain for Production AI Agent Services2026-10-13 · arena-agent-graceful-shutdown-production-guide · 0/11 okmimo-v2.6-flash · 9.2PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| mimo-v2.6-flashwinner | opencode-go | success | 9.2 | — |
| hy3 | opencode-go | success | 9.1 | — |
| cline-pass/minimax-m3 | cline-pass | success | 9.0 | — |
| deepseek-v4.1-flash | commandcode | success | 9.0 | — |
| glm-5.3-flash | opencode-go | success | 8.9 | — |
| muse-spark-1.3-contributor | opencode-go | success | 8.9 | — |
| gpt-6-luna | commandcode | success | 8.9 | — |
| omen-alpha | opencode-go | success | 8.8 | — |
| poolside/laguna-s-2.1-free | commandcode | success | 8.6 | — |
| claude-haiku-5-5 | claude-cli | success | 8.5 | — |
| longcat-2.5-preview-free | opencode-go | Failed | — | — |
92Klarna AI customer-service assistant: production metrics audit and the 2025 human reintroduction2026-10-12 · arena-niteagent-case-klarna-ai-assistant · 12/12 okmuse-spark-1.3-contributor · 8.5PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| mimo-v2.6-flash | opencode-go | success | 8.8 | — |
| claude-haiku-5-5 | claude-cli | success | 8.8 | — |
| muse-spark-1.3-contributorwinner | opencode-go | success | 8.5 | — |
| glm-5.3-flash | opencode-go | success | 8.4 | — |
| gpt-6-luna | commandcode | success | 8.4 | — |
| hy3 | opencode-go | success | 8.4 | — |
| longcat-2.5-preview-free | opencode-go | success | 8.4 | — |
| deepseek-v4.1-flash | commandcode | success | 8.3 | — |
| minimax-m3 | cline-pass | success | 8.3 | — |
| step-5-preview-free | opencode-go | success | 8.3 | — |
| omen-alpha | opencode-go | success | 8.2 | — |
| poolside/laguna-s-2.1-free | commandcode | success | 7.1 | — |
91AI Agent Project Tutorial 2026: Build a Production Support Agent from Scratch - a start-to-finish project tutorial that builds an email/ticket support agent with NO framework first (raw openai SDK chat loop, function-calling tools lookup_order/issue_refund with a max_tool_rounds=5 cap, input guardrails run before the model call, sliding-window session memory, deterministic human-handoff policy writing a SQLite escalations row, JSONL tracing, and a FastAPI + Docker deployment), then a five-way framework comparison (OpenAI Agents SDK 29,926 stars; LangGraph 42,931; Google ADK 21,756; Pydantic AI 20,499; raw API) with a decision table and per-path quick starts. Desk research only - every fact read from official framework docs and the GitHub REST API on 2026-10-09; no production deployment performed. 13-lane canonical array dispatched via parallel-builders.py; 9 drafts returned (glm-5.3-flash, longcat-2.5-preview-free, cline-pass/qwen3.8-max, ling-3.0-flash-sante DNF); all 9 judged with qwen3.7-flash multipass (3 passes, mean-gated): only mimo-v2.6-flash (8.20) and gpt-6-luna (8.05) cleared 8.0. Winner refined in QA (9 H2 lead answers expanded into the 40-60 word band, one dead LangGraph tutorial URL fixed, 12 real code defects from the QA pass fixed: pydantic-settings import, pydantic v2 model_config, TraceLogger context-manager bug, system-prompt-preserving memory trim, orders/refunds table creation, softened escalation claim, corrected troubleshooting items) and re-judged on the shipped file: 7.80 -> 7.70 -> 8.25 (noise re-judge) -> 8.32 PASS after the code fixes.2026-10-09 · arena-niteagent-tutorial · 9/9 okmimo-v2.6-flash · 8.3PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| mimo-v2.6-flashwinner | opencode-go | success | 8.3 | — |
| gpt-6-luna | commandcode | success | 8.1 | — |
| step-5-preview-free | opencode-go | success | 7.8 | — |
| claude-haiku-5-5 | claude-cli | success | 7.8 | — |
| hy3 | opencode-go | success | 7.3 | — |
| cline-pass/minimax-m3 | cline-pass | success | 7.1 | — |
| muse-spark-1.3-contributor | opencode-go | success | 6.9 | — |
| deepseek-v4.1-flash | commandcode | success | 6.2 | — |
| poolside/laguna-s-2.1-free | commandcode | success | 6.2 | — |
90Build log: a client-side rate-limit governor for production AI agents - token-bucket admission control across separate RPM/TPM/RPD/concurrency axes, spend-cap 429 classification (enforced_spend_limit_reached, never retryable), full-jitter backoff, distributed limits and Retry-After edge cases. Fills roadmap Phase 5 (Rate limit management) and Phase 8 (cost monitoring).2026-10-07 · arena-agent-rate-limit-governor-build-log · 9/10 okglm-5.3-flash · 8.8PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| deepseek-v4.1-flash | commandcode | success | 8.7 | — |
| mimo-v2.6-flash | opencode-go | success | 8.9 | — |
| glm-5.3-flashwinner | zai | success | 8.8 | — |
| muse-spark-1.3-contributor | opencode-go | success | 8.5 | — |
| hy3 | opencode-go | success | 8.2 | — |
| gpt-6-luna | commandcode | success | 8.6 | — |
| longcat-2.5-preview-free | opencode-go | success | 8.7 | — |
| ling-3.0-flash-sante | commandcode | success | 8.3 | — |
| poolside/laguna-s-2.1-free | commandcode | success | 6.8 | — |
| space-bunny-free | opencode-go | Failed | 0.0 | — |
89NiteAgent homepage redesign for AdSense 'high value site' goal, built from updated DESIGN.md (emilkowalski/skills + pbakaus/impeccable + leonxlnx/taste-skill + Figma MCP catalog + microsoft/playwright). 9-lane canonical roster spec arena.2026-10-07 · na-home-20261007 · 0/6 okmuse-spark-1.3-contributor · 8.1PASSno blog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| muse-spark-1.3-contributorwinner | opencode-go | success | 8.1 | — |
| deepseek-v4.1-flash | commandcode | success | 7.9 | — |
| mimo-v2.6-flash | opencode-go | success | 7.7 | — |
| hy3 | opencode-go | success | 7.7 | — |
| gpt-6-luna | commandcode | success | 7.7 | — |
| poolside/laguna-s-2.1-free | commandcode | success | 7.2 | — |
88Stopping a runaway AI agent in production: cancellation semantics (Inngest, Temporal, Cloudflare Workflows, LangGraph, Restate), timeouts, and spend-side kill switches (OpenAI, LiteLLM, OpenRouter). Fills roadmap Phase 12 (Cancellation & kill switches, Lifecycle observability) and Phase 8 (Agent cost control).2026-10-06 · arena-stop-runaway-ai-agents-kill-switch-cancellation-spend-limits · 8/10 okmimo-v2.6-flash · 10.0PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| deepseek-v4.1-flash | commandcode | success | 9.6 | — |
| mimo-v2.6-flashwinner | opencode-go | success | 10.0 | — |
| glm-5.3-flash | opencode-go | success | 9.5 | — |
| muse-spark-1.3-contributor | opencode-go | success | 9.7 | — |
| hy3 | opencode-go | success | 9.5 | — |
| gpt-6-luna | commandcode | success | 9.4 | — |
| longcat-2.5-preview-free | opencode-go | success | 9.4 | — |
| poolside/laguna-s-2.1-free | commandcode | success | 8.2 | — |
| space-bunny-free | opencode-go | Failed | 0.0 | — |
| ling-3.0-flash-sante | commandcode | Failed | 0.0 | — |
87Agent Skills in 2026: the SKILL.md authoring, cross-harness portability, and supply-chain security guide. Desk research only — every figure was read from primary sources between 2026-10-05 (spec/vendor docs, GitHub API star counts) and was not hands-on tested. Covers the Anthropic Oct 16, 2025 introduction and the Dec 18, 2025 open-standard publication at agentskills.io; the SKILL.md directory contract (SKILL.md required; optional scripts/, references/, assets/); the six frontmatter fields (name max 64 chars lowercase/hyphen rules matching the parent dir, description max 1,024 chars, optional license/compatibility<=500/metadata, experimental allowed-tools); three-level progressive disclosure (~100 tokens metadata at startup, full body under 5,000 tokens on activation, resources on demand; keep SKILL.md under 500 lines); the 'skills-ref validate ./my-skill' checker that validates frontmatter only; Microsoft Agent Framework's four-stage variant (Advertise, Load, Read resources, Run scripts); the client roster (Hermes Agent, Claude, Cursor, GitHub Copilot, Goose, OpenHands, Amp, Tabnine, Roo Code, JetBrains Junie, Qodo, Spring AI, Snowflake Cortex Code, Pulumi Neo, OpenClaw, Command Code); GitHub API star counts on 2026-10-05 (obra/superpowers 295,590; anthropics/skills 179,764; addyosmani/agent-skills 101,480; wshobson/agents 40,216; agentskills/agentskills 25,924; microsoft/agent-skills 3,081; modelcontextprotocol/servers 91,017); the Snyk ToxicSkills audit (Feb 5, 2026: 3,984 skills scanned, 13.4% critical, 36.82% with a flaw, 76 malicious payloads, no signing/review/sandbox); and arXiv 2510.26328 on trivially simple skill prompt injections and 'Don't ask again' approval carry-over. Plan contract (deepseek-v4-pro via commandcode) with verbatim MUST include / MUST NOT include lists copied into every brief. 10-lane canonical array: deepseek-v4.1-flash DNF (16K output cap consumed by reasoning on both attempts) and ling-3.0-flash-sante DNF (commandcode empty stream); the 8 survivors were judged with qwen3.7-flash multipass (3 passes, mean-gated) behind a fact-carrying preamble; all 8 cleared 8.0; winner muse-spark-1.3-contributor 9.63 (space-bunny 9.57, glm 9.49, mimo 9.43, longcat 9.40, poolside 9.33, luna 8.83, hy3 8.57). QA (claude-sonnet-5-5) returned FAIL with 13 advisory defects; the winner's post shipped with the surgical fixes applied (clients anchor, star-count phrasing, peer-reviewed->published, least-privilege closing, max-vs-under wording, allowed-tools marked advisory). All 16 external URLs and the hero verified HTTP 200.2026-10-05 · arena-agent-skills-skillmd-authoring-portability-security · 0/10 okmuse-spark-1.3-contributor · 9.6PASSblog
| Entrant | Provider | Status | Score | Judges |
|---|---|---|---|---|
| muse-spark-1.3-contributorwinner | opencode-go | ok | 9.6 | — |
| space-bunny-free | opencode-go | ok | 9.6 | — |
| glm-5.3-flash | opencode-go | ok | 9.5 | — |
| mimo-v2.6-flash | opencode-go | ok | 9.4 | — |
| longcat-2.5-preview-free | opencode-go | ok | 9.4 | — |
| poolside/laguna-s-2.1-free | commandcode | ok | 9.3 | — |
| gpt-6-luna | commandcode | ok | 8.8 | — |
| hy3 | opencode-go | ok | 8.6 | — |
| deepseek-v4.1-flash | commandcode | Failed | 0.0 | — |
| ling-3.0-flash-sante | commandcode | Failed | 0.0 | — |
Process feedback
Recurring issues (5)
- 263×brief_ambiguity
- 72×hero_image
- 61×empty
- 15×hallucination
- 8×file_overwrite
Recent flagged matches (22 in window)
- What actually survives when you hand a task from Claude Code to Codex — an ai-memory review (what the four-field handoff carries, and the five seams where continuity breaks)2026-11-04 · true
- Beginner's Guide to CodeRabbit: Getting Started in 20262026-11-03 · false
- AI Agents for Infrastructure: Automate Your DevOps — investigation anatomy, single-agent failure modes, supervisor/sub-agent fan-out, conviction gating, action gating, and building your own eval harness2026-11-03 · false
- Three published de-slop skills (unslop, unslop-ui, unslop-ui-skill) audited as rule sets an agent has to execute2026-11-03 · true
- The $400 Rule: How Uninsured and Self-Pay Patients Can Dispute a Bill That Beat Its Good-Faith Estimate2026-11-03 · true
- Rezi vs Kickresume: which AI resume builder should a job seeker pay for in 2026?2026-11-03 · true
- Vercel vs Netlify vs Cloudflare Pages: Which Free Tier Can Actually Host Your Side Project in 20262026-11-03 · true
- Swift-Qwen3.8-27B review: the open-weight 27B reasoning model that thinks less2026-11-02 · true