Task Planning & Knowledge Search

Artificial Analysis public leaderboard

τ²-Bench

Tool-agent interaction benchmark results published by Artificial Analysis.

Published source

Metric

Pass rate

Results shown

440

Publisher

Artificial Analysis

Snapshot

Fetched 2026-08-22T12:55:16.136Z

440 of 440 models

Higher is better

JT-35B-Flash

GLM-5.2 (max)

GLM-4.7-Flash

GLM-5-Turbo

GLM 5V Turbo

Step 3.7 Flash

Claude Fable 5 (with fallback)

GLM-5

GLM-5.1

Grok 4.3 (high)

Qwen3.6 Plus

GLM-5

GLM-5.1

Grok 4.20 0309

DeepSeek V4 Pro (max)

Kimi K2.6

Qwen3.6 Max Preview

GLM-4.7

Kimi K2.5

Qwen3.5 397B A17B

Gemini 3.5 Flash (medium)

DeepSeek V4 Flash (high)

Gemini 3.1 Pro Preview

Gemini 3.5 Flash

MiniMax-M2.5

Qwen3.6 35B A3B

MiMo-V2-Pro

MiMo-V2-Flash

DeepSeek V4 Flash (max)

Qwen3.7 Max

Claude Opus 4.8 (max)

DeepSeek V4 Flash

Step 3.5 Flash

MiMo-V2.5-Pro

DeepSeek V4 Pro (high)

GLM-4.7

Qwen3.6 27B

Mistral Medium 3.5

Kimi K2.6

Qwen3.5 27B

GPT-5.5 (xhigh)

Qwen3.5 122B A10B

Qwen3.6 27B

MiMo-V2-Flash (Feb 2026)

Grok 4.1 Fast

Tri-21B-think Preview

JT-MINI

Qwen3.7 Plus

GPT-5.5 (high)

Kimi K2 Thinking

Grok 4.20 0309 v2

Nova 2.0 Pro Preview (medium)

Hy3-preview

Ring-2.6-1T

GPT-5.2 Codex (xhigh)

Claude Opus 4.6 (max)

Qwen3.5 4B

GPT-5.5 (medium)

GLM-4.7-Flash

Muse Spark

Grok 4.3 (medium)

DeepSeek V4 Pro

MiMo-V2-Omni

MiMo-V2.5

DeepSeek V3.2

Nova 2.0 Pro Preview (low)

Grok 3 mini Reasoning (high)

Kimi K2.7 Code

Trinity Large Thinking

Ling-2.6-1T

Claude Opus 4.5

KAT-Coder-Pro V2

Qwen3.5 35B A3B

MiniMax-M3

Grok 4.3 (low)

Claude Opus 4.7 (max)

KAT-Coder-Pro V1

Qwen3.5 Omni Plus

MiMo-V2-Omni-0327

Qwen3.5 4B

MiniCPM-V 4.6 1.3B

Step 3.5 Flash 2603

HyperCLOVA X SEED Think (32B)

Qwen3.5 27B

GPT-5.4 (xhigh)

Gemini 3 Pro Preview (high)

Qwen3.5 9B

GPT-5 Codex (high)

MiniMax-M2

GPT-5 (medium)

Mi:dm K 2.5 Pro

Claude Opus 4.5

Solar Pro 3

GPT-5.6 Terra (max)

Qwen3.5 35B A3B

GPT-5.3 Codex (xhigh)

Ling 2.6 Flash

MiniMax-M2.1

Qwen3.5 9B

Qwen3.6 35B A3B

GPT-5.6 Sol (max)

Claude Opus 4.6 (high)

GPT-5 (high)

GPT-5.2 (xhigh)

MiniMax-M2.7

GPT-5.6 Sol (xhigh)

Qwen3.5 122B A10B

Qwen3.5 Omni Flash

GPT-5 (low)

Qwen3.5 397B A17B

MiMo-V2-Flash

GPT-5.5 (low)

ERNIE 5.0 Thinking Preview

Qwen3 Max Thinking

Qwen3 Max Thinking (Preview)

Nemotron 3 Ultra

GPT-5.6 Sol (high)

GPT-5.4 mini (xhigh)

GPT-5.1 Codex (high)

MiniCPM5-1B

GPT-5.1 (high)

Qwen3.5 2B

Nex-N2-Pro

Kimi K2.5

Tri-21B-Think

GPT-5.6 Sol (medium)

MiniCPM5-1B

Command A+

Nova 2.0 Omni (medium)

Gemini 3 Flash

GPT-5.6 Terra (xhigh)

Claude Sonnet 4.6 (Non-reasoning)

LongCat Flash Lite

Qwen3 Coder Next

DeepSeek V3.2

Claude Sonnet 4.6 (Non-reasoning, Low Effort)

GPT-5.6 Terra (high)

EXAONE 4.5 33B

Claude 4.5 Sonnet

GLM-4.6

GPT-5.6 Sol (low)

GPT-5.4 nano (xhigh)

Grok Code Fast 1

Claude Sonnet 4.6 (max)

Nova 2.0 Lite (medium)

Grok 4

GPT-5.4 (low)

K-

K-EXAONE

Qwen3 Max

GPT-5.2 (medium)

Claude Opus 4.7 (Non-reasoning, high)

Kimi K2 0905

Claude 4 Opus

GPT-5.6 Terra (medium)

Nova 2.0 Lite (high)

MiMo-V2.5-Pro

Nova 2.0 Lite (low)

Nova 2.0 Pro Preview

Claude 4.1 Opus

GPT-5 mini (medium)

Mercury 2

GLM-4.6

Claude 4.5 Sonnet

Grok 4.20 0309

GPT-5.5 (Non-reasoning)

Apriel-v1.6-15B-Thinker

Qwen3.5 2B

GPT-5 mini (high)

Apriel-v1.5-15B-Thinker

Gemini 3 Pro Preview (low)

Nemotron 3 Super

Nova 2.0 Omni (low)

Hy3-preview

GPT-5 (minimal)

Grok 4 Fast

Grok 4.3 (Non-reasoning)

gpt-oss-120b (high)

Gemma 4 31B

Qwen3.5 0.8B

Claude 4 Sonnet

Grok 4 Fast

Grok 4.1 Fast

HyperNova 60B 2605

GPT-5.1 Codex mini (high)

Nova 2.0 Lite

Kimi K2

GPT-5.6 Terra (low)

gpt-oss-20b (high)

Grok 4.20 0309 v2

Gemma 4 31B

K-

K-EXAONE

Gemini 3.5 Flash (minimal)

Doubao Seed Code

o4-mini (high)

Claude 3.7 Sonnet

Claude 4.5 Haiku

Gemini 2.5 Pro

Qwen3 VL 235B A22B

Nemotron Cascade 2 30B A3B

Qwen3 235B A22B 2507

GPT-4.1 mini

GPT-5.4 nano

Claude 4 Sonnet

Magistral Medium 1.2

gpt-oss-20b (low)

Claude 3.7 Sonnet

Seed-OSS-36B-Instruct

Mi:dm K 2.5 Pro Preview

GPT-5.5 Instant (May 2026)

Grok 3

Solar Open 100B

Qwen3.5 0.8B

GPT-4.1

DeepSeek V3 0324

Sarvam 105B (high)

GLM-4.5-Air

Motif-2-12.7B

GPT-5.2

GPT-5.1

Gemini 2.5 Flash (Sep)

Qwen3 VL 32B

Nemotron 3 Nano Omni 30B A3B

gpt-oss-120b (low)

Nova 2.0 Omni

Qwen3 Coder 480B

Gemma 4 26B A4B

Gemini 3 Flash

GLM-4.5

Granite 4.1 30B

Qwen3 Next 80B A3B

Mistral Small 4

Nemotron 3 Nano

Mistral Medium 3.1

Gemma 4 26B A4B

Nova Premier

Devstral Small (May)

DeepSeek V3.1

North Mini Code

DeepSeek V3.1 Terminus

GPT-5.4 mini (medium)

DeepSeek R1 0528

Pixtral Large

GPT-5 nano (high)

Gemma 4 12B

GPT-5.4 (Non-reasoning)

Qwen3 VL 235B A22B

GPT-5.4 nano

DeepSeek V3.1

Qwen3 14B

Sarvam 30B (high)

Qwen2.5 72B

Qwen3 Coder 30B A3B

MiniMax M1 80k

DeepSeek V3.2 Exp

Qwen3 235B 2507

Mistral Large 2 (Jul)

Ling-1T

Qwen3 Max (Preview)

Claude 4.5 Haiku

Qwen3 14B

Gemma 4 12B (Non-reasoning)

Solar Pro 2

GPT-5 mini (minimal)

Gemini 2.5 Flash

MiniMax M1 40k

GLM-4.6V

o3-mini (high)

Gemini 3.1 Flash-Lite

GLM-4.6V

Mistral Large 2 (Nov)

Gemini 2.5 Flash-Lite (Sep)

GPT-5 nano (medium)

Qwen3 32B

Mistral Small 3.2

Gemini 2.0 Flash

Qwen3 VL 8B

Qwen3 VL 32B

GPT-4o (Aug)

o3-mini

Gemini 2.5 Flash (Sep)

Devstral Small

Qwen3 30B A3B 2507

Nemotron 3 Nano 4B

Llama Nemotron Super 49B v1.5

Solar Pro 2

Magistral Small 1.2

Granite 4.1 8B

Qwen3 8B

K2-V2 (high)

Falcon-H1R-7B

Qwen3 235B

Ministral 3 14B

Llama 3.3 Nemotron Super 49B

Magistral Small 1

INTELLECT-3

Ministral 3 8B

Llama 3.3 70B

Qwen3 4B 2507

Hermes 4 405B

Ring-1T

Qwen3 30B

Gemma 4 E4B

Qwen3 1.7B

GPT-5 nano (minimal)

Nemotron 3 Nano

Qwen3 4B 2507

K2 Think V2

Mistral Small 3.1

Llama Nemotron Super 49B v1.5

GPT-4o (Nov)

Devstral 2

Ministral 3 3B

Qwen3 8B

K2-V2 (medium)

Claude 3.5 Haiku

Mistral Large 3

Mistral Medium 3

Qwen3 235B

NVIDIA Nemotron Nano 9B V2

GPT-5.4 mini

Devstral Small 2

Qwen3 VL 4B

Magistral Medium 1

Llama 3.1 Nemotron 70B

DeepSeek V3 (Dec)

Granite 4.0 1B

GLM-4.5V

Hermes 4 70B

Qwen3 VL 8B

Gemma 4 E2B

Hermes 4 405B

Qwen3 30B

DeepSeek R1 Distill Llama 70B

NVIDIA Nemotron Nano 9B V2

Hermes 4 70B

Qwen3 Next 80B A3B

Qwen3 1.7B

Nanbeige4.1-3B

NVIDIA Nemotron Nano 12B v2 VL

Olmo 3.1 32B Instruct

Qwen3 Omni 30B A3B

Claude 3 Haiku

Qwen3 0.6B

Llama 3.2 3B

Ling-flash-2.0

Gemma 4 E2B

Gemma 4 E4B

K2-V2 (low)

Exaone 4.0 1.2B

Solar Mini

Devstral Medium

Qwen3 VL 30B A3B

GLM-4.5V

LFM2.5-1.2B-Thinking

Granite 4.1 3B

Granite 4.0 H 1B

Mistral Small 3

NVIDIA Nemotron Nano 12B v2 VL

Qwen3 4B

Gemini 2.5 Flash-Lite

Llama 3.1 405B

Qwen3 VL 30B A3B

Mistral Small 4

Gemini 2.5 Flash-Lite

Llama 4 Maverick

Nova Lite

EXAONE 4.0 32B

Granite 4.0 H Small

GPT-4.1 nano

Qwen3 Omni 30B A3B

Llama 3.1 8B

Exaone 4.0 1.2B

Step3 VL 10B

LFM2.5-8B-A1B

Jamba Reasoning 3B

Llama 4 Scout

Qwen3 VL 4B

Command A

Llama 3.1 70B

Gemini 2.5 Flash

Qwen3 0.6B

Llama 3.2 11B (Vision)

Granite 4.0 H 350M

Nova Micro

Nova Pro

LFM2 2.6B

Jamba 1.7 Large

Ling-mini-2.0

Granite 4.0 350M

Apertus 70B Instruct

LFM2 1.2B

Granite 4.0 Micro

Olmo 3 7B

Jamba 1.7 Mini

Llama 3.1 Nemotron Nano 4B v1.1

DeepSeek R1 (Jan)

Apertus 8B Instruct

Llama Nemotron Ultra

LFM2 24B A2B

Gemma 3 12B

LFM2.5-1.2B-Instruct

Gemma 3 27B

Granite 3.3 8B

LFM2 8B A1B

Gemma 3 1B

Qwen3 30B A3B 2507

Gemma 3 270M

LFM2.5-VL-1.6B

Phi-4 Mini

Gemma 3 4B

Gemma 3n E4B

EXAONE 4.0 32B

Mistral 7B

Olmo 3 7B Think

Sarvam M

Olmo 3.1 32B Think

Molmo2-8B

Ring-flash-2.0

DeepSeek V3.2 Speciale

OLMo 2 32B

Kimi Linear 48B A3B Instruct

DeepSeek R1 0528 Qwen3 8B

Llama 3.2 1B

Tiny Aya Global

Llama 3 8B

Gemma 3n E2B

Llama 3 70B

ERNIE 4.5 300B A47B

Molmo 7B-D

Olmo 3 32B Think

Reka Flash 3

Phi-3 Mini

GPT-5 (ChatGPT)

Phi-4

OLMo 2 7B

List bars visualize relative performance. Plot bars use published 0–100 scores, shrink to a 1.5rem minimum, then scroll horizontally.

Scores are published snapshots and are only comparable within the same benchmark, version, metric, and evaluation configuration.