Recommendation for OpenClaw
OpenClaw
Our top recommendation for OpenClaw, based on the public evidence we track, is OpenAI: GPT-5.6 Sol. Anthropic: Claude Fable 5 is the next-ranked alternative.
About this recommendation
- Updated
- Sep 7, 2026
- Evidence through
- Sep 7, 2026
- Sources
- 5
- Revision
- v52
Decision audit
Why this result
Inspect the inputs and the computed order behind the recommendation.
Models screened
20
live candidates
Evaluation feeds
6
task-weighted
Winner coverage
79%
intended feed weight
Largest provider share
3 of 6
Anthropic
Sources evaluated
The task sets these weights before any model is scored.
| Evaluation feed | Weight | Winner result | Field measured |
|---|---|---|---|
| Berkeley Function Calling | 32% | not measured | 3/20 |
| LMArena Agent | 23% | #3 | 17/20 |
| price weight | 15% | 99/100 | 20/20 |
| Terminal-Bench 2.1 | 15% | #14 | 11/20 |
| SWE-rebench | 10% | #4 | 15/20 |
| OpenRouter usage | 5% | 97/100 | 20/20 |
Provider concentration
Each exact model is scored separately; provider identity is not a ranking input.
- Anthropic3 models
- OpenAI2 models
- Z.ai1 model
Decision table
Every published model is shown in computed order. Practitioner sources are distinct community threads, not the citations repeated in the prose below.
| Rank | Model | Relative score | Coverage | Practitioner evidence | Strongest measured reason |
|---|---|---|---|---|---|
| 01 | GPT-5.6 SolOpenAI | 80 | 79% | 3 threads · 2 families · 1 cautions | #3 LMArena Agent · #4 SWE-rebench |
| 02 | Claude Fable 5Anthropic | 72 | 79% | no linked practitioner threads | #1 SWE-rebench · #2 LMArena Agent |
| 03 | GLM 5.2Z.ai | 70 | 73% | no linked practitioner threads | #8 LMArena Agent · #9 SWE-rebench |
| 04 | GPT-5.6 LunaOpenAI | 67 | 79% | 2 threads · 2 families · 1 cautions | #4 Terminal-Bench 2.1 · #18 LMArena Agent |
| 05 | Claude Opus 4.6Anthropic | 66 | 73% | no linked practitioner threads | #15 LMArena Agent · #16 SWE-rebench |
| 06 | Claude Sonnet 4.6Anthropic | 65 | 73% | no linked practitioner threads | #13 SWE-rebench · #21 LMArena Agent |
Relative score combines normalized benchmark quality and signal coverage; independent practitioner evidence and freshness are bounded tie-breakers. It is an ordering score, not an absolute quality percentage. The writing model receives this order and cannot change it.
OpenAI: GPT-5.6 Sol ranks #3 of 47 on LMArena's agentic arena (score 9.3), measuring tool-use and multi-step task performance from human preference.
Best when: Consider only after reviewing the cited caution.
Anthropic: Claude Fable 5 ranks #2 of 47 on LMArena's agentic arena (score 10.2), measuring tool-use and multi-step task performance from human preference.
Best when: Consider only after reviewing the cited caution.
GLM 5.2 is the highest-ranked open-weight option on LMArena's agentic arena, placing tenth among 47 models for tool-use and multi-step tasks.
Best when: Consider only after reviewing the cited caution.
Watch out for
- Expect lower task completion rates compared to top closed-weight alternatives, as its 6.0 agentic arena score trails the leaders by a significant margin.
GPT-5.6 Luna ranks near the bottom of LMArena's agentic arena and shows elevated failure rates on code patch verification in local audits.
Best when: Consider only after reviewing the cited caution.
Watch out for
- Avoid for implementation-heavy OpenClaw skills involving code patches, with local audits showing 239 failed `apply_patch` calls including 229 with explicit 'Failed' errors.
- Expect the same Responses Lite integration complexity as other GPT-5.6 variants, requiring client-side `web.run` extensions instead of hosted tools.
Claude Opus 4.6 sits mid-table on LMArena's agentic arena with a 4.6 score, underperforming relative to Claude Fable 5 on multi-step tool-use tasks.
Best when: Consider only after reviewing the cited caution.
Watch out for
- Consider skipping for pure agentic workloads where Fable 5 is available, as its #16 ranking and 4.6 score indicate weaker human preference for tool-use performance.
Claude Sonnet 4.6 ranks second-to-last on LMArena's agentic arena with a minimal 0.5 score for tool-use and multi-step task performance.
Best when: Consider only after reviewing the cited caution.
Watch out for
- Steer clear of long-horizon agent deployments, as its 0.5 agentic arena score and #23 ranking suggest poor reliability on multi-step tool-calling sequences.
Sources
- 1
“Ranks #10 of 47 on LMArena's agentic arena (score 6), measuring tool-use and multi-step task performance from human preference.”
LMArena agentic arena · Benchmark · Sep 5, 2026 - 2
“## Summary The OpenAI preset currently sets `openai/gpt-5.6-luna` as default for both implementation-heavy lanes: - Designer: `medium` - Fixer: `xhigh` Local runs show a high number of `apply_patch` verification failures while using Luna. Could documentation and preset guidance be reviewed, with an alternative default or an explicit caveat for patch-heavy work? ## Local audit Observed across local OpenCode session history: - `gpt-5.6-luna`: 239 failed `apply_patch` calls - 229 contained `Failed…”
ricatix · GitHub · Jul 28, 2026 - 3
“## Environment - sub2api: current `main` (`f18f3143e` during reproduction) - Codex CLI: `0.145.0` - Client transport: custom API-key Responses provider pointing at sub2api - Codex config: `web_search = "live"` - Affected models: `gpt-5.6-sol`, `gpt-5.6-terra`, `gpt-5.6-luna` ## Problem The Codex model catalog marks the three GPT-5.6 models above with `use_responses_lite=true`. In Responses Lite mode, Codex removes hosted Responses tools and expects a client-side `web.run` extension instead. For…”
Lang-code-ops · GitHub · Jul 27, 2026 - 4
“Ranks #16 of 47 on LMArena's agentic arena (score 4.6), measuring tool-use and multi-step task performance from human preference.”
LMArena agentic arena · Benchmark · Sep 5, 2026 - 5
“Ranks #23 of 47 on LMArena's agentic arena (score 0.5), measuring tool-use and multi-step task performance from human preference.”
LMArena agentic arena · Benchmark · Sep 5, 2026
Rankings synthesized from community evidence and open benchmarks. See methodology. Not driven by vendor marketing.