Part of Best LLM for AI Agents
Best LLM for Tool & Function Calling
Claude Fable 5.1 is the provisional first pick for Tool & Function Calling as of October 20, 2026, while independent task measurements are still limited.[1][2][3][4][5] It ranks #1 of 48 on LMArena Agent's tool-hallucination signal (score 0.7), where a higher score means fewer fabricated tool calls. GPT-6 Astra scores within two points, so read the top two as close rather than as a measured gap. GPT-6 Astra is the next-ranked alternative. Ranks #2 of 48 on LMArena's agentic arena (score 15.1) and steerability (score 15.5), tied for #1 on tool-hallucination signal (score 0.7).
About this recommendation
- Updated
- Evidence through
- Sources
- 17
- Revision
- v111
Decision audit
Why this result
Inspect the inputs and the computed order behind the recommendation.
Models screened
24
live candidates
Evaluation feeds
7
task-weighted
Winner coverage
46%
intended feed weight
Largest provider share
3 of 8
Anthropic
Anthropic: Claude Fable 5.1 leads the measured task comparison with limited independent task measurements. Community volume, release age and usage popularity add no ranking points. The leading scores are close under this policy.
Real-world track record is still developing. The recommendation can lead on independent measurements while practitioner reports accumulate.
Sources evaluated
These are the weights used in scoring. Missing measurements are omitted from each model’s average. A feed with no result for any model released in the last 60 days is stale and does not score until it catches up.
| Evaluation feed | Weight | Winner result | Field measured |
|---|---|---|---|
| Berkeley Function Callingstale | 0% | no current result | 0/24 |
| tau2-benchstale | 0% | no current result | 0/24 |
| LMArena Agent Tool HallucinationObserved 2026-10-19 · evaluation date not supplied · max effort | 36% | #1 | 24/24 |
| LMArena AgentObserved 2026-10-19 · evaluation date not supplied · max effort | 27% | #3 | 24/24 |
| LMArena Agent SteerabilityObserved 2026-10-19 · evaluation date not supplied · max effort | 18% | #3 | 24/24 |
| LMArena SearchObserved 2026-10-13 · evaluation date not supplied · claude-fable-5.1-high | 18% | #2 | 4/24 |
| Structured-output evaldisabled; last successful check unknownunavailable | 0% | feed unavailable | 0/24 |
Provider concentration
Each exact model is scored separately; provider identity is not a ranking input.
- Anthropic3 models
- Meta2 models
- OpenAI2 models
- Google1 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.
Anthropic · coverage 46% · provisional evidence
#1 LMArena Agent Tool Hallucination · #2 LMArena Search
OpenAI · coverage 46% · close score · provisional evidence
#1 LMArena Agent Tool Hallucination · #2 LMArena Agent
Anthropic · coverage 46% · close score · provisional evidence
#1 LMArena Agent · #1 LMArena Agent Steerability
OpenAI · coverage 46% · provisional evidence
#1 LMArena Agent Tool Hallucination · #4 LMArena Agent
Google · coverage 38% · provisional evidence
#3 LMArena Agent Steerability · #6 LMArena Agent
Anthropic · coverage 38% · provisional evidence
#5 LMArena Agent · #7 LMArena Agent Steerability
Meta · coverage 38% · provisional evidence
#10 LMArena Agent Tool Hallucination · #15 LMArena Agent
Meta · coverage 38% · provisional evidence
#12 LMArena Agent · #17 LMArena Agent Steerability
Task score averages the available, task-weighted measurements. Benchmark positions compare models matched to our catalog. Models measured by at least two independent evaluator families are considered first. Missing tests, practitioner report counts, popularity and release age add no points. Coverage describes evidence breadth, not a probability of correctness. Scores within two points are shown as close under this policy; this is not a statistical confidence interval. When the leading scores are that close, the model measured by more evaluator families ranks first.
- 01
Claude Fable 5.1Anthropic · $10 in / $50 out per M tokens · 1M context
Claude Fable 5.1 ranks #1 of 48 on LMArena Agent's tool-hallucination signal (score 0.7), where a higher score means fewer fabricated tool calls.
Best when: Its strongest result for this task is on LMArena Agent tool hallucination, which measures calling only tools that exist, with real arguments.No practitioner reports are linked yet, so no cautions are listed. Absence of reports is not evidence of reliability.
- 02
GPT-6 AstraOpenAI · $10 in / $50 out per M tokens · 1.1M context
Ranks #2 of 48 on LMArena's agentic arena (score 15.1) and steerability (score 15.5), tied for #1 on tool-hallucination signal (score 0.7).
Best when: Use for low-hallucination tool calling where fabricated calls must be minimized, as it ties for the best tool-hallucination signal score (0.7) in the evaluation.More tips
- Deploy when mid-task direction following is critical, ranking #2 on steerability (score 15.5) among 48 models.
Watch out for
- Expect sensitivity to AGENTS.md instructions and skills, with tendencies toward detailed formatting, more clarifying questions, and over-testing small changes.
- 03
Claude Opus 5.5Anthropic · $4 in / $20 out per M tokens · 1M context
Ranks #1 of 48 on LMArena's agentic arena (score 20.4) for tool-use and multi-step task performance, and #1 on steerability (score 17.3).
Best when: Use for complex multi-step agent workflows requiring mid-task direction changes, as it leads all models in steerability (score 17.3) and agentic arena performance (score 20.4).More tips
- Deploy for subagent-heavy applications like document processing pipelines, demonstrated in building a complete 387-label map translation app with subagents in one afternoon.
- 04
GPT-6.1 SolOpenAI · $2 in / $10 out per M tokens · 1.1M context
GPT-6.1 Sol ranks #1 of 48 on LMArena Agent's tool-hallucination signal (score 0.7), where a higher score means fewer fabricated tool calls.
Best when: Its strongest result for this task is on LMArena Agent tool hallucination, which measures calling only tools that exist, with real arguments. - 05
Gemini 4 ArgonGoogle · $2 in / $10 out per M tokens · 1.0M context
Ranks #6 of 48 on LMArena's agentic arena (score 12.1) and #3 on steerability (score 14.8).
Best when: Use when mid-task direction following is prioritized, as it ranks #3 on steerability (score 14.8).Watch out for
- Expect frequent fabricated tool calls, ranking #30 of 48 on tool-hallucination signal (score 0.1).
- 06
Claude Sonnet 5.5Anthropic · $2 in / $10 out per M tokens · 1M context
Claude Sonnet 5.5 ranks #5 of 48 on LMArena's agentic arena (score 12.8), measuring tool-use and multi-step task performance from human preference.
Best when: Its strongest result for this task is on LMArena Agent, which measures tool calling and multi-step control across a long task. - 07
Muse Spark 1.3Meta · $1.25 in / $4.25 out per M tokens · 1.0M context
Muse Spark 1.3 ranks #10 of 48 on LMArena Agent's tool-hallucination signal (score 0.6), where a higher score means fewer fabricated tool calls.
Best when: Its strongest result for this task is on LMArena Agent tool hallucination, which measures calling only tools that exist, with real arguments.No practitioner reports are linked yet, so no cautions are listed. Absence of reports is not evidence of reliability.
- 08
Muse WonderMeta · $4 in / $20 out per M tokens · 1.0M context
Ranks #12 of 48 on LMArena's agentic arena (score 3.8) but shows poor tool-hallucination performance (#28 of 48, score 0.2) and steerability (#17 of 48, score 4.8), with limited available evidence.
Best when: Its strongest result for this task is on LMArena Agent, which measures tool calling and multi-step control across a long task.Watch out for
- Expect higher rates of fabricated tool calls, ranking #28 of 48 on tool-hallucination signal (score 0.2).
- Watch for limited mid-task direction following, ranking #17 of 48 on steerability (score 4.8).
Frequently asked
- What is the best model for Tool & Function Calling?
- Claude Fable 5.1 ranks first as of October 20, 2026. Claude Fable 5.1 ranks #1 of 48 on LMArena Agent's tool-hallucination signal (score 0.7), where a higher score means fewer fabricated tool calls.[1]
Sources
- 1
“Within the 48 catalog models matched to this evaluation: Ranks #1 of 48 on LMArena Agent's tool-hallucination signal (score 0.7), where a higher score means fewer fabricated tool calls.”
LMArena Agent tool hallucination · Benchmark · Oct 19, 2026 - 2
“Within the 48 catalog models matched to this evaluation: Ranks #2 of 48 on LMArena's agentic arena (score 15.1), measuring tool-use and multi-step task performance from human preference.”
LMArena agentic arena · Benchmark · Oct 19, 2026 - 3
“Within the 48 catalog models matched to this evaluation: Ranks #2 of 48 on LMArena Agent's steerability signal (score 15.5), measuring how well the agent follows direction given mid-task.”
LMArena Agent steerability · Benchmark · Oct 19, 2026 - 4
“Within the 48 catalog models matched to this evaluation: Ranks #1 of 48 on LMArena Agent's tool-hallucination signal (score 0.7), where a higher score means fewer fabricated tool calls.”
LMArena Agent tool hallucination · Benchmark · Oct 19, 2026 - 5
“Summary Two distinct tool-calling failures were found using the NEAR AI provider from OpenCode (which uses the OpenAI-compatible /v1/chat/completions endpoint). 1. openai/gpt-6-astra — function tools rejected on /v1/chat/completions OpenAI rejects function tools for this model on /v1/chat/completions; tools only work on /v1/responses. OpenCode's openai-compatible provider uses /v1/chat/completions, so agentic runs error out. Not fixable from OpenCode config. Update to:…”
think-in-universe · GitHub · Sep 17, 2026 - 6
“Context GPT-6 Astra is available, but its token usage currently makes it unsuitable as the default day-to-day model. Use it selectively as a high-capability workflow and prompt auditor. Official guidance notes that Astra is more sensitive to instructions in AGENTS.md and skills, may ask more clarifying questions, tends toward detailed formatting, and can over-test small changes. These behaviors overlap with prior dotfiles friction around unnecessary STOPs, repeated confirmations, verbose w…”
jjshen-eland · GitHub · Sep 18, 2026 - 7
“Within the 48 catalog models matched to this evaluation: Ranks #1 of 48 on LMArena Agent's steerability signal (score 17.3), measuring how well the agent follows direction given mid-task.”
LMArena Agent steerability · Benchmark · Oct 19, 2026 - 8
“Within the 48 catalog models matched to this evaluation: Ranks #1 of 48 on LMArena's agentic arena (score 20.4), measuring tool-use and multi-step task performance from human preference.”
LMArena agentic arena · Benchmark · Oct 19, 2026 - 9
“How a map on a café wall turned into an app that reads, maps and translates all 387 labels of an 1819 French map of the Bosphorus, built with Claude Opus 5.5 subagents in an afternoon. I wrote up where this came from and how I built it: https: cahidarda.com articles bosphore-1819”
cahid_arda · Hacker News · Sep 24, 2026 - 10
“Within the 48 catalog models matched to this evaluation: Ranks #1 of 48 on LMArena Agent's tool-hallucination signal (score 0.7), where a higher score means fewer fabricated tool calls.”
LMArena Agent tool hallucination · Benchmark · Oct 19, 2026 - 11
“Within the 48 catalog models matched to this evaluation: Ranks #3 of 48 on LMArena Agent's steerability signal (score 14.8), measuring how well the agent follows direction given mid-task.”
LMArena Agent steerability · Benchmark · Oct 19, 2026 - 12
“Within the 48 catalog models matched to this evaluation: Ranks #30 of 48 on LMArena Agent's tool-hallucination signal (score 0.1), where a higher score means fewer fabricated tool calls.”
LMArena Agent tool hallucination · Benchmark · Oct 19, 2026 - 13
“Within the 48 catalog models matched to this evaluation: Ranks #5 of 48 on LMArena's agentic arena (score 12.8), measuring tool-use and multi-step task performance from human preference.”
LMArena agentic arena · Benchmark · Oct 19, 2026 - 14
“Within the 48 catalog models matched to this evaluation: Ranks #10 of 48 on LMArena Agent's tool-hallucination signal (score 0.6), where a higher score means fewer fabricated tool calls.”
LMArena Agent tool hallucination · Benchmark · Oct 19, 2026 - 15
“Within the 48 catalog models matched to this evaluation: Ranks #12 of 48 on LMArena's agentic arena (score 3.8), measuring tool-use and multi-step task performance from human preference.”
LMArena agentic arena · Benchmark · Oct 19, 2026 - 16
“Within the 48 catalog models matched to this evaluation: Ranks #28 of 48 on LMArena Agent's tool-hallucination signal (score 0.2), where a higher score means fewer fabricated tool calls.”
LMArena Agent tool hallucination · Benchmark · Oct 19, 2026 - 17
“Within the 48 catalog models matched to this evaluation: Ranks #17 of 48 on LMArena Agent's steerability signal (score 4.8), measuring how well the agent follows direction given mid-task.”
LMArena Agent steerability · Benchmark · Oct 19, 2026
Rankings synthesized from community evidence and open benchmarks. See methodology. Not driven by vendor marketing.