τ²-Bench Telecom
Customer-support tool use requiring an agent and user to coordinate actions in a shared telecom environment.
Tool useArtificial Analysis323 models run by Artificial AnalysisUpdated 30 September 2026
How it is run
Artificial Analysis publishes its independent telecom-domain runs. This table selects the highest published effort per model release, retaining the setting.
What it cannot tell you
Telecom results are domain-specific and separate from τ³-Banking; scores from different domains or versions must not be merged.
- Leader
- JT-35B-Flash99.12% · Artificial Analysis results, grouped here by release
- Top-ten spread
- 1.46%Leader minus tenth place, or last place in smaller fields.
- Median
- 36.55%of the 323 rows published
- No published result
- 36of 46 models on this site
Artificial Analysis results by release
Highest published effort per model release, grouped here from AA’s public results. Grouping ranks are derived here; source settings and model-publisher reports stay separate.
| # | Model | Score | Effort |
|---|---|---|---|
| 1 | JT-35B-Flash | 99.12% | Not specified |
| 1 | GLM-5.2 | 99.12% | max |
| 3 | GLM-4.7-Flash | 98.83% | Not specified |
| 4 | GLM-5-Turbo | 98.54% | Not specified |
| 4 | GLM 5V Turbo | 98.54% | Not specified |
| 4 | Step 3.7 Flash | 98.54% | high |
| 4 | Claude Fable 5 | 98.54% | max |
| 8 | GLM-5 | 98.25% | Not specified |
| 9 | GLM-5.1 | 97.66% | Not specified |
| 9 | Grok 4.3 | 97.66% | high |
| 9 | Qwen3.6 Plus | 97.66% | Not specified |
| 12 | Grok 4.20 0309 | 96.49% | Not specified |
| 13 | DeepSeek V4 Pro 0424 | 96.20% | max |
| 14 | Kimi K2.6 | 95.91% | Not specified |
| 14 | Qwen3.6 Max Preview | 95.91% | Not specified |
| 14 | GLM-4.7 | 95.91% | Not specified |
| 14 | Kimi K2.5 | 95.91% | Not specified |
| 18 | Qwen3.5 397B A17B | 95.61% | Not specified |
| 18 | Gemini 3.1 Pro Preview | 95.61% | Not specified |
| 20 | Gemini 3.5 Flash | 95.32% | high |
| 20 | MiniMax-M2.5 | 95.32% | Not specified |
| 22 | MiMo-V2-Pro | 95.03% | Not specified |
| 22 | DeepSeek V4 Flash 0420 | 95.03% | max |
| 24 | Qwen3.7 Max | 94.74% | Not specified |
| 25 | Claude Opus 4.8 | 94.44% | max |
| 25 | Step 3.5 Flash | 94.44% | Not specified |
| 27 | MiMo-V2.5-Pro | 94.15% | Not specified |
| 27 | Mistral Medium 3.5 | 94.15% | high |
| 29 | Qwen3.5 27B | 93.86% | Not specified |
| 29 | GPT-5.5 | 93.86% | xhigh |
| 31 | Qwen3.6 27B | 93.57% | Not specified |
| 32 | MiMo-V2-Flash (Feb 2026) | 93.27% | Not specified |
| 32 | Grok 4.1 Fast | 93.27% | Not specified |
| 32 | Tri-21B-think Preview | 93.27% | Not specified |
| 35 | JT-MINI | 92.98% | Not specified |
| 35 | Qwen3.7 Plus | 92.98% | Not specified |
| 35 | Kimi K2 Thinking | 92.98% | Not specified |
| 38 | Nova 2.0 Pro Preview | 92.69% | medium |
| 38 | Hy3-preview | 92.69% | Not specified |
| 40 | Ring-2.6-1T | 92.40% | xhigh |
| 41 | GPT-5.2 Codex | 92.11% | xhigh |
| 41 | Claude Opus 4.6 | 92.11% | max |
| 43 | Muse Spark | 91.52% | Not specified |
| 44 | MiMo-V2-Omni | 91.23% | Not specified |
| 45 | MiMo-V2.5 | 90.64% | Not specified |
| 46 | Grok 3 mini Reasoning | 90.35% | high |
| 47 | Kimi K2.7 Code | 90.06% | Not specified |
| 47 | Trinity Large Thinking | 90.06% | Not specified |
| 49 | Ling-2.6-1T | 89.77% | Not specified |
| 50 | Claude Opus 4.5 | 89.47% | Not specified |
| 50 | KAT Coder Pro V2 | 89.47% | Not specified |
| 52 | Qwen3.5 35B A3B | 89.18% | Not specified |
| 53 | MiniMax-M3 | 88.89% | Not specified |
| 54 | Claude Opus 4.7 | 88.60% | max |
| 54 | KAT-Coder-Pro V1 | 88.60% | Not specified |
| 56 | Qwen3.5 Omni Plus | 88.30% | Not specified |
| 57 | MiMo-V2-Omni-0327 | 88.01% | Not specified |
| 58 | Qwen3.5 4B | 87.72% | Not specified |
| 58 | MiniCPM-V 4.6 1.3B | 87.72% | Not specified |
| 60 | Step 3.5 Flash 2603 | 87.43% | Not specified |
| 60 | HyperCLOVA X SEED Think (32B) | 87.43% | Not specified |
| 62 | GPT-5.4 | 87.13% | xhigh |
| 62 | Gemini 3 Pro Preview | 87.13% | high |
| 64 | GPT-5 Codex | 86.84% | high |
| 64 | MiniMax-M2 | 86.84% | Not specified |
| 66 | Mi:dm K 2.5 Pro | 86.55% | Not specified |
| 67 | GPT-5.6 Terra | 86.26% | max |
| 67 | Solar Pro 3 | 86.26% | high |
| 69 | GPT-5.3 Codex | 85.96% | xhigh |
| 69 | Ling 2.6 Flash | 85.96% | Not specified |
| 71 | MiniMax-M2.1 | 85.38% | Not specified |
| 72 | GPT-5.6 Sol | 85.09% | max |
| 72 | Qwen3.5 9B | 85.09% | Not specified |
| 72 | Qwen3.6 35B A3B | 85.09% | Not specified |
| 75 | GPT-5 | 84.80% | high |
| 75 | GPT-5.2 | 84.80% | xhigh |
| 75 | MiniMax-M2.7 | 84.80% | Not specified |
| 78 | Qwen3.5 122B A10B | 84.50% | Not specified |
| 78 | Qwen3.5 Omni Flash | 84.50% | Not specified |
| 80 | MiMo-V2-Flash | 83.92% | Not specified |
| 80 | ERNIE 5.0 Thinking Preview | 83.92% | Not specified |
| 82 | Qwen3 Max Thinking | 83.63% | Not specified |
| 82 | Qwen3 Max Thinking (Preview) | 83.63% | Not specified |
| 84 | GPT-5.4 mini | 83.33% | xhigh |
| 84 | Nemotron 3 Ultra 550B A55B | 83.33% | Not specified |
| 86 | GPT-5.1 Codex | 83.04% | high |
| 87 | GPT-5.1 | 81.87% | high |
| 88 | Qwen3.5 2B | 81.58% | Not specified |
| 88 | Nex-N2-Pro | 81.58% | Not specified |
| 90 | Tri-21B-Think | 80.99% | Not specified |
| 90 | MiniCPM5-1B | 80.99% | Not specified |
| 92 | o3 | 80.70% | Not specified |
| 92 | Command A+ | 80.70% | Not specified |
| 94 | Nova 2.0 Omni | 80.41% | medium |
| 95 | LongCat Flash Lite | 79.53% | Not specified |
| 95 | Qwen3 Coder Next | 79.53% | Not specified |
| 97 | DeepSeek V3.2 | 78.95% | Not specified |
| 98 | EXAONE 4.5 33B | 78.07% | Not specified |
| 98 | Claude 4.5 Sonnet | 78.07% | Not specified |
| 100 | GPT-5.4 nano | 76.02% | xhigh |
| 101 | Claude Sonnet 4.6 | 75.73% | max |
| 101 | Grok Code Fast 1 | 75.73% | Not specified |
| 103 | Grok 4 | 74.85% | Not specified |
| 104 | K-EXAONE | 74.27% | Not specified |
| 104 | Qwen3 Max | 74.27% | Not specified |
| 106 | Kimi K2 0905 | 73.39% | Not specified |
| 106 | Claude 4 Opus | 73.39% | Not specified |
| 108 | Nova 2.0 Lite | 72.81% | high |
| 109 | Claude 4.1 Opus | 71.37% | Not specified |
| 110 | Mercury 2 | 70.76% | high |
| 111 | GLM-4.6 | 70.47% | Not specified |
| 112 | Apriel-v1.6-15B-Thinker | 69.30% | Not specified |
| 113 | GPT-5 mini | 68.42% | high |
| 113 | Apriel-v1.5-15B-Thinker | 68.42% | Not specified |
| 115 | Nemotron 3 Super 120B A12B | 67.84% | Not specified |
| 116 | gpt-oss-120b | 65.79% | high |
| 117 | Qwen3.5 0.8B | 65.20% | Not specified |
| 118 | Claude 4 Sonnet | 64.62% | Not specified |
| 119 | Grok 4 Fast | 63.74% | Not specified |
| 120 | HyperNova 60B 2605 | 63.16% | high |
| 121 | GPT-5.1 Codex mini | 62.87% | high |
| 122 | o1 | 62.57% | Not specified |
| 123 | Kimi K2 | 61.11% | Not specified |
| 124 | gpt-oss-20b | 60.23% | high |
| 125 | Grok 4.20 0309 v2 | 59.94% | Not specified |
| 125 | Gemma 4 31B | 59.94% | Not specified |
| 127 | Doubao Seed Code | 58.19% | Not specified |
| 128 | o4-mini | 55.56% | high |
| 129 | Claude 3.7 Sonnet | 54.68% | Not specified |
| 129 | Claude 4.5 Haiku | 54.68% | Not specified |
| 131 | Gemini 2.5 Pro | 54.09% | Not specified |
| 131 | Qwen3 VL 235B A22B (Reasoning) | 54.09% | Not specified |
| 133 | Nemotron Cascade 2 30B A3B | 53.22% | Not specified |
| 133 | Qwen3 235B A22B 2507 (Reasoning) | 53.22% | Not specified |
| 135 | GPT-4.1 mini | 52.92% | Not specified |
| 136 | Magistral Medium 1.2 | 52.05% | Not specified |
| 137 | Seed-OSS-36B-Instruct | 49.42% | Not specified |
| 137 | Mi:dm K 2.5 Pro Preview | 49.42% | Not specified |
| 137 | GPT-5.5 Instant (May 2026) | 49.42% | Not specified |
| 140 | Grok 3 | 48.83% | Not specified |
| 141 | Solar Open 100B | 48.25% | high |
| 142 | GPT-4.1 | 47.08% | Not specified |
| 142 | DeepSeek V3 0324 | 47.08% | Not specified |
| 144 | Sarvam 105B | 46.78% | high |
| 145 | GLM-4.5-Air | 46.49% | Not specified |
| 145 | Motif-2-12.7B-Reasoning | 46.49% | Not specified |
| 147 | Qwen3 VL 32B (Reasoning) | 45.61% | Not specified |
| 148 | Nemotron 3 Nano Omni 30B A3B Reasoning | 45.32% | Not specified |
| 149 | Qwen3 Coder 480B A35B Instruct | 43.57% | Not specified |
| 149 | Gemma 4 26B A4B | 43.57% | Not specified |
| 151 | Gemini 3 Flash Preview | 43.27% | Not specified |
| 152 | GLM-4.5 | 42.98% | Not specified |
| 153 | Granite 4.1 30B | 42.11% | Not specified |
| 154 | Qwen3 Next 80B A3B (Reasoning) | 41.52% | Not specified |
| 155 | Mistral Medium 3.1 | 40.64% | Not specified |
| 156 | Nova Premier | 38.30% | Not specified |
| 157 | Devstral Small (May '25) | 38.01% | Not specified |
| 158 | DeepSeek V3.1 | 37.43% | Not specified |
| 158 | North Mini Code | 37.43% | Not specified |
| 160 | DeepSeek V3.1 Terminus | 37.13% | Not specified |
| 161 | GPT-5 nano | 36.55% | high |
| 161 | DeepSeek R1 0528 (May '25) | 36.55% | Not specified |
| 161 | Pixtral Large | 36.55% | Not specified |
| 164 | Qwen3 VL 235B A22B Instruct | 35.09% | Not specified |
| 165 | Qwen3 14B | 34.50% | Not specified |
| 165 | Sarvam 30B | 34.50% | high |
| 165 | Qwen2.5 Instruct 72B | 34.50% | Not specified |
| 165 | Qwen3 Coder 30B A3B Instruct | 34.50% | Not specified |
| 169 | MiniMax M1 80k | 34.21% | Not specified |
| 170 | DeepSeek V3.2 Exp | 33.92% | Not specified |
| 171 | Qwen3 235B A22B 2507 | 33.33% | Not specified |
| 172 | Mistral Large 2 (Jul '24) | 33.04% | Not specified |
| 173 | Ling-1T | 32.75% | Not specified |
| 173 | Qwen3 Max (Preview) | 32.75% | Not specified |
| 175 | Gemma 4 12B | 31.87% | Not specified |
| 176 | Gemini 2.5 Flash | 31.58% | Not specified |
| 176 | MiniMax M1 40k | 31.58% | Not specified |
| 178 | o3-mini | 31.29% | high |
| 178 | Gemini 3.1 Flash-Lite | 31.29% | high |
| 180 | GLM-4.6V | 30.70% | Not specified |
| 180 | Mistral Large 2 (Nov '24) | 30.70% | Not specified |
| 182 | Gemini 2.5 Flash-Lite Preview (Sep '25) | 30.41% | Not specified |
| 183 | Qwen3 32B | 29.82% | Not specified |
| 184 | Mistral Small 3.2 | 29.53% | Not specified |
| 184 | Gemini 2.0 Flash (Feb '25) | 29.53% | Not specified |
| 186 | Qwen3 VL 8B Instruct | 29.24% | Not specified |
| 186 | Qwen3 VL 32B Instruct | 29.24% | Not specified |
| 188 | GPT-4o (Aug '24) | 28.95% | Not specified |
| 189 | Gemini 2.5 Flash Preview (Sep '25) | 28.36% | Not specified |
| 189 | Devstral Small (Jul '25) | 28.36% | Not specified |
| 191 | Solar Pro 2 | 28.07% | high |
| 191 | Qwen3 30B A3B 2507 (Reasoning) | 28.07% | Not specified |
| 191 | NVIDIA Nemotron 3 Nano 4B | 28.07% | Not specified |
| 194 | Magistral Small 1.2 | 27.78% | Not specified |
| 194 | Granite 4.1 8B | 27.78% | Not specified |
| 194 | Qwen3 8B | 27.78% | Not specified |
| 194 | K2-V2 | 27.78% | high |
| 194 | Falcon-H1R-7B | 27.78% | Not specified |
| 199 | Qwen3 235B A22B | 27.19% | Not specified |
| 199 | Ministral 3 14B | 27.19% | Not specified |
| 201 | Llama 3.3 Nemotron Super 49B v1 | 26.90% | Not specified |
| 202 | Magistral Small 1 | 26.61% | Not specified |
| 202 | INTELLECT-3 | 26.61% | Not specified |
| 202 | Ministral 3 8B | 26.61% | Not specified |
| 202 | Llama 3.3 Instruct 70B | 26.61% | Not specified |
| 202 | Qwen3 4B 2507 | 26.61% | Not specified |
| 207 | Ring-1T | 26.32% | Not specified |
| 208 | Qwen3 30B A3B | 26.02% | Not specified |
| 208 | Gemma 4 E4B | 26.02% | Not specified |
| 210 | NVIDIA Nemotron 3 Nano 30B A3B | 25.44% | Not specified |
| 210 | Qwen3 4B 2507 (Reasoning) | 25.44% | Not specified |
| 210 | K2 Think V2 | 25.44% | high |
| 213 | Mistral Small 3.1 | 25.15% | Not specified |
| 213 | Llama Nemotron Super 49B v1.5 | 25.15% | Not specified |
| 213 | GPT-4o (Nov '24) | 25.15% | Not specified |
| 216 | Devstral 2 | 24.85% | Not specified |
| 216 | Ministral 3 3B | 24.85% | Not specified |
| 218 | Claude 3.5 Haiku | 24.56% | Not specified |
| 218 | Mistral Large 3 | 24.56% | Not specified |
| 220 | Mistral Medium 3 | 24.27% | Not specified |
| 221 | NVIDIA Nemotron Nano 9B V2 | 23.39% | Not specified |
| 221 | Devstral Small 2 | 23.39% | Not specified |
| 221 | Qwen3 VL 4B Instruct | 23.39% | Not specified |
| 224 | Magistral Medium 1 | 23.10% | Not specified |
| 224 | Llama 3.1 Nemotron Instruct 70B | 23.10% | Not specified |
| 226 | DeepSeek V3 (Dec '24) | 22.81% | Not specified |
| 226 | Granite 4.0 1B | 22.81% | Not specified |
| 228 | Qwen3 VL 8B (Reasoning) | 22.51% | Not specified |
| 229 | Gemma 4 E2B | 22.22% | Not specified |
| 229 | Hermes 4 - Llama-3.1 405B | 22.22% | Not specified |
| 231 | DeepSeek R1 Distill Llama 70B | 21.93% | Not specified |
| 232 | Hermes 4 - Llama-3.1 70B | 21.64% | Not specified |
| 232 | Qwen3 Next 80B A3B Instruct | 21.64% | Not specified |
| 232 | Qwen3 1.7B | 21.64% | Not specified |
| 232 | Nanbeige4.1-3B | 21.64% | Not specified |
| 236 | NVIDIA Nemotron Nano 12B v2 VL | 21.35% | Not specified |
| 236 | Olmo 3.1 32B Instruct | 21.35% | Not specified |
| 236 | Qwen3 Omni 30B A3B (Reasoning) | 21.35% | Not specified |
| 239 | Claude 3 Haiku | 21.05% | Not specified |
| 239 | Llama 3.2 Instruct 3B | 21.05% | Not specified |
| 241 | Ling-flash-2.0 | 20.76% | Not specified |
| 242 | Exaone 4.0 1.2B | 20.47% | Not specified |
| 243 | Solar Mini | 20.18% | Not specified |
| 244 | Devstral Medium | 19.88% | Not specified |
| 244 | Qwen3 VL 30B A3B (Reasoning) | 19.88% | Not specified |
| 246 | GLM-4.5V | 19.59% | Not specified |
| 246 | LFM2.5-1.2B-Thinking | 19.59% | Not specified |
| 246 | Granite 4.1 3B | 19.59% | Not specified |
| 246 | Granite 4.0 H 1B | 19.59% | Not specified |
| 246 | Mistral Small 3 | 19.59% | Not specified |
| 251 | Qwen3 4B | 19.01% | Not specified |
| 251 | Gemini 2.5 Flash-Lite | 19.01% | Not specified |
| 251 | Llama 3.1 Instruct 405B | 19.01% | Not specified |
| 251 | Qwen3 VL 30B A3B Instruct | 19.01% | Not specified |
| 255 | Mistral Small 4 | 18.42% | Not specified |
| 256 | Llama 4 Maverick | 17.84% | Not specified |
| 257 | Nova Lite | 17.54% | Not specified |
| 258 | Granite 4.0 H Small | 17.25% | Not specified |
| 258 | GPT-4.1 nano | 17.25% | Not specified |
| 260 | Qwen3 Omni 30B A3B Instruct | 16.37% | Not specified |
| 260 | Llama 3.1 Instruct 8B | 16.37% | Not specified |
| 262 | Step3 VL 10B | 16.08% | Not specified |
| 262 | LFM2.5-8B-A1B | 16.08% | Not specified |
| 264 | Jamba Reasoning 3B | 15.79% | Not specified |
| 265 | Llama 4 Scout | 15.50% | Not specified |
| 265 | Qwen3 VL 4B (Reasoning) | 15.50% | Not specified |
| 267 | Command A | 15.20% | Not specified |
| 267 | Llama 3.1 Instruct 70B | 15.20% | Not specified |
| 269 | Qwen3 0.6B | 14.62% | Not specified |
| 269 | Llama 3.2 Instruct 11B (Vision) | 14.62% | Not specified |
| 269 | Granite 4.0 H 350M | 14.62% | Not specified |
| 272 | Nova Micro | 14.04% | Not specified |
| 272 | Nova Pro | 14.04% | Not specified |
| 274 | LFM2 2.6B | 13.45% | Not specified |
| 274 | Jamba 1.7 Large | 13.45% | Not specified |
| 276 | Ling-mini-2.0 | 13.16% | Not specified |
| 276 | Granite 4.0 350M | 13.16% | Not specified |
| 278 | Apertus 70B Instruct | 12.87% | Not specified |
| 279 | LFM2 1.2B | 12.57% | Not specified |
| 279 | Granite 4.0 Micro | 12.57% | Not specified |
| 279 | Olmo 3 7B Instruct | 12.57% | Not specified |
| 279 | Jamba 1.7 Mini | 12.57% | Not specified |
| 283 | Llama 3.1 Nemotron Nano 4B v1.1 | 11.70% | Not specified |
| 284 | DeepSeek R1 (Jan '25) | 11.40% | Not specified |
| 284 | Apertus 8B Instruct | 11.40% | Not specified |
| 284 | Llama 3.1 Nemotron Ultra 253B v1 | 11.40% | Not specified |
| 287 | LFM2 24B A2B | 11.11% | Not specified |
| 288 | Gemma 3 12B Instruct | 10.82% | Not specified |
| 288 | LFM2.5-1.2B-Instruct | 10.82% | Not specified |
| 290 | Gemma 3 27B Instruct | 10.53% | Not specified |
| 290 | Granite 3.3 8B | 10.53% | Not specified |
| 290 | LFM2 8B A1B | 10.53% | Not specified |
| 290 | Gemma 3 1B Instruct | 10.53% | Not specified |
| 294 | Qwen3 30B A3B 2507 | 10.23% | Not specified |
| 295 | Gemma 3 270M | 9.06% | Not specified |
| 296 | LFM2.5-VL-1.6B | 8.48% | Not specified |
| 297 | Phi-4 Mini Instruct | 8.19% | Not specified |
| 298 | Gemma 3 4B Instruct | 4.97% | Not specified |
| 298 | Gemma 3n E4B Instruct | 4.97% | Not specified |
| 300 | EXAONE 4.0 32B | 4.09% | Not specified |
| 301 | Mistral 7B Instruct | 0.00% | Not specified |
| 301 | Olmo 3 7B Think | 0.00% | Not specified |
| 301 | Sarvam M | 0.00% | high |
| 301 | Olmo 3.1 32B Think | 0.00% | Not specified |
| 301 | Molmo2-8B | 0.00% | Not specified |
| 301 | Ring-flash-2.0 | 0.00% | Not specified |
| 301 | DeepSeek V3.2 Speciale | 0.00% | Not specified |
| 301 | OLMo 2 32B | 0.00% | Not specified |
| 301 | Kimi Linear 48B A3B Instruct | 0.00% | Not specified |
| 301 | DeepSeek R1 0528 Qwen3 8B | 0.00% | Not specified |
| 301 | Llama 3.2 Instruct 1B | 0.00% | Not specified |
| 301 | Tiny Aya Global | 0.00% | Not specified |
| 301 | Llama 3 Instruct 8B | 0.00% | Not specified |
| 301 | Gemma 3n E2B Instruct | 0.00% | Not specified |
| 301 | Llama 3 Instruct 70B | 0.00% | Not specified |
| 301 | ERNIE 4.5 300B A47B | 0.00% | Not specified |
| 301 | Molmo 7B-D | 0.00% | Not specified |
| 301 | Olmo 3 32B Think | 0.00% | Not specified |
| 301 | Reka Flash 3 | 0.00% | Not specified |
| 301 | Phi-3 Mini Instruct 3.8B | 0.00% | Not specified |
| 301 | GPT-5 (ChatGPT) | 0.00% | Not specified |
| 301 | Phi-4 | 0.00% | Not specified |
| 301 | OLMo 2 7B | 0.00% | Not specified |
No published result tracked
36 of the 46 models on this site have no published number on this benchmark, so their cells in the home table stay blank rather than reading as zero.
Step 5 PreviewGPT-6 LunaGLM 5.3 FlashQwen3.8 2.4T A95BQwen3.8-Flash-NextMiMo-V2.6-FlashQwen3.8 Max (0902)Grok 4.6Gemini 3.7 FlashMuse Spark 1.2Grok 4.5DeepSeek V4.1 Flash 1017GLM-LegalClaude Haiku 5.5BeamGemini 4 ArgonClaude Sonnet 5.5Claude Opus 5.5Claude Fable 5.1Claude Opus 5Claude Sonnet 5GPT-6 AstraGPT-6 SolGPT-6.1 SolGPT-5.6 LunaGemini 3.8 FlashGrok 4.7DeepSeek V4.1 FlashDeepSeek V4 ProKimi K3Qwen3.8 MaxGLM 5.3Muse Spark 1.3 MaxHunyuan Hy4 PreviewMiMo V2.6 ProInkling
Read this with
Benchmarks in the same category, which usually disagree with this one on purpose.
Sources
Every page used for this benchmark, read on 17 October 2026.
- Benchmark site or the page that publishes itartificialanalysis.ai/evaluations/tau2-bench
- Independent resultsartificialanalysis.ai/evaluations/tau2-bench