The daily briefing

GPT-5.6 Sol: OpenAI publishes coding evaluations for low-resource languages

Tuesday, October 20, 2026 · what moved in the AI release cycle in those 24 hours.

One click creates your free account (or signs you into an existing one) and turns on alerts for the Daily Briefing. There's no password to set, and you can unsubscribe any time.

New benchmark readings

Amazon Nova 2 Pro GDP.pdf: 2 · Epoch AI
Qwen3.8 LM Arena: 1482.3530183862679 · LM Arena
Qwen3.8 Coding: 53.53 · BenchLM.ai
Qwen3.8 Agentic: 67.45 · BenchLM.ai
Qwen3.8 Reasoning: 81.3 · BenchLM.ai
Qwen3.8 Knowledge: 68.15 · BenchLM.ai
Qwen3.8 MultimodalGrounded: 91.2 · BenchLM.ai
Qwen3.8 InstructionFollowing: 94.7 · BenchLM.ai
Amazon Nova Humanity's Last Exam: 4.4 · Epoch AI
Amazon Nova 2 Lite LM Arena: 1334.7838251387504 · LM Arena
Claude 3 Haiku LM Arena: 1261.7688839616312 · LM Arena
Claude 3 Opus LM Arena: 1322.2806893188556 · LM Arena
Claude 3 Sonnet LM Arena: 1281.449412868066 · LM Arena
Claude 3.5 Haiku LM Arena: 1324.7789160325767 · LM Arena
Claude 3.5 Sonnet Humanity's Last Exam: 4.1 · Epoch AI
Claude 3.5 Sonnet LM Arena: 1374.2560571692657 · LM Arena
Claude 3.7 Sonnet Humanity's Last Exam: 8 · Epoch AI
Claude 3.7 Sonnet LM Arena: 1372.568246809387 · LM Arena
Claude Fable 5 LM Arena: 1503.6013443401307 · LM Arena
Claude Fable 5 Coding: 71.33 · BenchLM.ai
Claude Fable 5 Agentic: 75.84 · BenchLM.ai
Claude Fable 5 GDP.pdf: 30 · Epoch AI
Claude Fable 5 Knowledge: 82.59 · BenchLM.ai
Claude Fable 5.1 Coding: 76.87 · BenchLM.ai
Claude Fable 5.1 Agentic: 80.78 · BenchLM.ai
Claude Fable 5.1 Reasoning: 85.2 · BenchLM.ai
Claude Fable 5.1 Knowledge: 86.87 · BenchLM.ai
Claude Fable 5.1 Humanity's Last Exam: 46.5 · Epoch AI
Claude Fable 5.1 GDP.pdf: 29.6 · Epoch AI
Claude Fable 5.1 LM Arena: 1502.2862278899863 · LM Arena
Claude Haiku 4.5 ExploitBench: 13.7 · Epoch AI
Claude Haiku 4.5 LM Arena: 1414.264434778655 · LM Arena
Claude Haiku 5.5 Coding: 51.93 · BenchLM.ai
Claude Haiku 5.5 Agentic: 61.43 · BenchLM.ai
Claude Haiku 5.5 Reasoning: 60.8 · BenchLM.ai
Claude Haiku 5.5 Knowledge: 66.38 · BenchLM.ai
Claude Haiku 5.5 Humanity's Last Exam: 30.1 · Epoch AI
Claude Mythos 5 ExploitBench: 70.6 · Epoch AI
Claude Opus 4 Humanity's Last Exam: 10.7 · Epoch AI
Claude Opus 4 LM Arena: 1414.2718576179084 · LM Arena
Claude Opus 4.1 Humanity's Last Exam: 11.5 · Epoch AI
Claude Opus 4.1 LM Arena: 1448.08122844461 · LM Arena
Claude Opus 4.5 Math: 61.3 · BenchLM.ai
Claude Opus 4.5 MultimodalGrounded: 48.6 · BenchLM.ai
Claude Opus 4.5 Multilingual: 96.5 · BenchLM.ai
Claude Opus 4.5 Humanity's Last Exam: 25.2 · Epoch AI
Claude Opus 4.5 LM Arena: 1469.6308482689085 · LM Arena
Claude Opus 4.6 LM Arena: 1503.0532375944142 · LM Arena
Claude Opus 4.6 Coding: 49.07 · BenchLM.ai
Claude Opus 4.6 MultimodalGrounded: 74.5 · BenchLM.ai
Claude Opus 4.6 Humanity's Last Exam: 34.4 · Epoch AI
Claude Opus 4.7 LM Arena: 1501.0578036294407 · LM Arena
Claude Opus 4.7 Coding: 55.57 · BenchLM.ai
Claude Opus 4.7 Agentic: 59.73 · BenchLM.ai
Claude Opus 4.7 Knowledge: 66.17 · BenchLM.ai
Claude Opus 4.7 Humanity's Last Exam: 36.2 · Epoch AI
Claude Opus 4.7 ExploitBench: 26.5 · Epoch AI
Claude Opus 4.7 GDP.pdf: 21 · Epoch AI
Claude Opus 4.7 MultimodalGrounded: 53.7 · BenchLM.ai
Claude Opus 4.8 LM Arena: 1481.4953019365078 · LM Arena
Claude Opus 4.8 Coding: 60.7 · BenchLM.ai
Claude Opus 4.8 Agentic: 63.33 · BenchLM.ai
Claude Opus 4.8 Knowledge: 72 · BenchLM.ai
Claude Opus 4.8 MultimodalGrounded: 91.2 · BenchLM.ai
Claude Opus 4.8 GDP.pdf: 24 · Epoch AI
Claude Opus 4.8 Math: 84.4 · BenchLM.ai
Claude Opus 4.8 Reasoning: 62.8 · BenchLM.ai
Claude Opus 5 Coding: 70.83 · BenchLM.ai
Claude Opus 5 Agentic: 78.45 · BenchLM.ai
Claude Opus 5 Knowledge: 82.25 · BenchLM.ai
Claude Opus 5 MultimodalGrounded: 92.2 · BenchLM.ai
Claude Opus 5 Reasoning: 81 · BenchLM.ai
Claude Opus 5 GDP.pdf: 24 · Epoch AI
Claude Opus 5 LM Arena: 1489.9918464257669 · LM Arena
Claude Opus 5.5 LM Arena: 1508.1664796431178 · LM Arena
Claude Opus 5.5 Coding: 82.73 · BenchLM.ai
Claude Opus 5.5 Agentic: 90.37 · BenchLM.ai
Claude Opus 5.5 Reasoning: 86.1 · BenchLM.ai
Claude Opus 5.5 Knowledge: 90.23 · BenchLM.ai
Claude Opus 5.5 MultimodalGrounded: 92.2 · BenchLM.ai
Claude Opus 5.5 GDP.pdf: 30.6 · Epoch AI
Claude Sonnet 4 Humanity's Last Exam: 7.8 · Epoch AI
Claude Sonnet 4 LM Arena: 1390.8871111749847 · LM Arena
Claude Sonnet 4.5 Humanity's Last Exam: 13.7 · Epoch AI
Claude Sonnet 4.5 LM Arena: 1454.9317881127183 · LM Arena
Claude Sonnet 4.6 LM Arena: 1472.102664549635 · LM Arena
Claude Sonnet 4.6 MultimodalGrounded: 58.7 · BenchLM.ai
Claude Sonnet 4.6 ExploitBench: 23.6 · Epoch AI
Claude Sonnet 4.6 GDP.pdf: 18 · Epoch AI
Claude Sonnet 5 LM Arena: 1461.0744345634098 · LM Arena
Claude Sonnet 5 Coding: 57.51 · BenchLM.ai
Claude Sonnet 5 Agentic: 66.29 · BenchLM.ai
Claude Sonnet 5 Knowledge: 65.37 · BenchLM.ai
Claude Sonnet 5 MultimodalGrounded: 82 · BenchLM.ai
Command A LM Arena: 1353.7064642233104 · LM Arena
Command A InstructionFollowing: 90.7 · BenchLM.ai
Command R LM Arena: 1276.3819106946182 · LM Arena
DeepSeek LLM LM Arena: 1185.0649345408938 · LM Arena
DeepSeek V4 Flash LM Arena: 1435.6475480811423 · LM Arena
DeepSeek V4 Pro SWE-Bench Pro: 55.4 · Hugging Faceself reported
DeepSeek V4 Pro LM Arena: 1457.7921446281084 · LM Arena
DeepSeek-R1 LM Arena: 1421.4290222328934 · LM Arena
DeepSeek-V2.5 LM Arena: 1307.1758939412227 · LM Arena
DeepSeek-V3 LM Arena: 1395.194455554951 · LM Arena
DeepSeek-V3 Multilingual: 55.5 · BenchLM.ai
DeepSeek-V3.1 LM Arena: 1417.0286131400032 · LM Arena
DeepSeek-V3.2-Exp LM Arena: 1425.0893087677296 · LM Arena
ERNIE 5.1 LM Arena: 1467.0070290704896 · LM Arena
Gemini 1.0 LM Arena: 1223.7230219942562 · LM Arena
Gemini 1.5 Flash LM Arena: 1309.3034391516185 · LM Arena
Gemini 1.5 Pro Humanity's Last Exam: 4.6 · Epoch AI
Gemini 1.5 Pro LM Arena: 1351.3340444210469 · LM Arena
Gemini 2.0 Flash Humanity's Last Exam: 6.6 · Epoch AI
Gemini 2.0 Flash LM Arena: 1359.9029625115695 · LM Arena
Gemini 2.5 Flash Humanity's Last Exam: 12.1 · Epoch AI
Gemini 2.5 Flash LM Arena: 1409.2369734409983 · LM Arena
Gemini 2.5 Pro Humanity's Last Exam: 21.6 · Epoch AI
Gemini 2.5 Pro LM Arena: 1445.413742345976 · LM Arena
Gemini 3 Pro Knowledge: 60.86 · BenchLM.ai
Gemini 3 Pro MultimodalGrounded: 80 · BenchLM.ai
Gemini 3 Pro InstructionFollowing: 86.1 · BenchLM.ai
Gemini 3 Pro Humanity's Last Exam: 37.5 · Epoch AI
Gemini 3 Pro LM Arena: 1485.4590679512698 · LM Arena
Gemini 3.1 Pro Knowledge: 67.68 · BenchLM.ai
Gemini 3.1 Pro MultimodalGrounded: 80.1 · BenchLM.ai
Gemini 3.1 Pro GDP.pdf: 17 · Epoch AI
Gemini 3.1 Pro Humanity's Last Exam: 46.4 · Epoch AI
Gemini 3.1 Pro ExploitBench: 26.1 · Epoch AI
Gemini 3.1 Pro LM Arena: 1486.9094358879568 · LM Arena
Gemini 3.5 Flash LM Arena: 1477.4750000563504 · LM Arena
Gemini 3.5 Flash Coding: 50.06 · BenchLM.ai
Gemini 3.5 Flash Agentic: 51.31 · BenchLM.ai
Gemini 3.5 Flash Reasoning: 70.9 · BenchLM.ai
Gemini 3.5 Flash MultimodalGrounded: 87.8 · BenchLM.ai
Gemini 3.5 Flash InstructionFollowing: 87.4 · BenchLM.ai
Gemini 3.5 Flash GDP.pdf: 14 · Epoch AI
Gemini 3.5 Flash Knowledge: 65.49 · BenchLM.ai
Gemini 3.5 Flash-Lite LM Arena: 1454.7574024087119 · LM Arena
Gemini 3.5 Flash-Lite GDP.pdf: 10 · Epoch AI
Gemini 3.6 Flash LM Arena: 1482.7020039923352 · LM Arena
Gemini 3.6 Flash Coding: 51.04 · BenchLM.ai
Gemini 3.6 Flash Knowledge: 64.84 · BenchLM.ai
Gemini 3.6 Flash GDP.pdf: 14 · Epoch AI
Gemini 3.7 Flash Coding: 57.63 · BenchLM.ai
Gemini 3.7 Flash Agentic: 59.21 · BenchLM.ai
Gemini 3.7 Flash Knowledge: 71.34 · BenchLM.ai
Gemini 3.7 Flash MultimodalGrounded: 87.2 · BenchLM.ai
Gemini 3.7 Flash LM Arena: 1487.5878753892953 · LM Arena
Gemini 4 Argon Coding: 73.07 · BenchLM.ai
Gemini 4 Argon Agentic: 78.71 · BenchLM.ai
Gemini 4 Argon Knowledge: 81.39 · BenchLM.ai
Gemini 4 Argon LM Arena: 1523.9035861299108 · LM Arena
Gemini 4 Argon Reasoning: 86.9 · BenchLM.ai
Glm 5 2 Coding: 53.68 · BenchLM.ai
Glm 5 2 Agentic: 57.4 · BenchLM.ai
Glm 5 2 InstructionFollowing: 89.9 · BenchLM.ai
Glm 5 2 Math: 78.4 · BenchLM.ai
GLM-4 LM Arena: 1273.4793154674621 · LM Arena
GLM-4.5 Humanity's Last Exam: 8.3 · Epoch AI
GLM-4.5 LM Arena: 1410.7960767851844 · LM Arena
GLM-4.6 SWE-Bench Pro: 9.7 · Hugging Faceself reported
GLM-4.6 LM Arena: 1424.1228406450284 · LM Arena
GLM-4.7 LM Arena: 1441.3266291777354 · LM Arena
GLM-5 Multilingual: 89.5 · BenchLM.ai
GLM-5 InstructionFollowing: 88.6 · BenchLM.ai
GLM-5 LM Arena: 1457.2666631824843 · LM Arena
GLM-5.1 ExploitBench: 16 · Epoch AI
GLM-5.1 SWE-Bench Pro: 58.4 · Hugging Faceself reported
GLM-5.1 LM Arena: 1465.1654788552419 · LM Arena
GLM-5.2 LM Arena: 1475.8811079028058 · LM Arena
GLM-5.2 SWE-Bench Pro: 62.1 · Hugging Faceself reported
GPT-3.5 Turbo LM Arena: 1225.7069421076292 · LM Arena
GPT-4 LM Arena: 1276.3446463311536 · LM Arena
GPT-4 Turbo LM Arena: 1324.5281327823136 · LM Arena
GPT-4.1 Humanity's Last Exam: 5.4 · Epoch AI
GPT-4.1 LM Arena: 1414.5863221306088 · LM Arena
GPT-4.1 Multilingual: 61.5 · BenchLM.ai
GPT-4.5 Humanity's Last Exam: 5.4 · Epoch AI
GPT-4.5 LM Arena: 1444.6270945007745 · LM Arena
GPT-4o Humanity's Last Exam: 2.7 · Epoch AI
GPT-4o LM Arena: 1346.1327118570525 · LM Arena
GPT-4o Multilingual: 30.2 · BenchLM.ai
GPT-4o mini LM Arena: 1317.5796589492732 · LM Arena
GPT-5 InstructionFollowing: 86.4 · BenchLM.ai
GPT-5 Humanity's Last Exam: 31.6 · Epoch AI
GPT-5 LM Arena: 1434.8035075319092 · LM Arena
GPT-5.1 InstructionFollowing: 89.4 · BenchLM.ai
GPT-5.1 Humanity's Last Exam: 23.7 · Epoch AI
GPT-5.1 LM Arena: 1455.7154324566077 · LM Arena
GPT-5.2 Humanity's Last Exam: 27.8 · Epoch AI
GPT-5.2 LM Arena: 1437.2456671445352 · LM Arena
GPT-5.4 Coding: 47.75 · BenchLM.ai
GPT-5.4 Agentic: 52.2 · BenchLM.ai
GPT-5.4 Reasoning: 64.2 · BenchLM.ai
GPT-5.4 Knowledge: 69.16 · BenchLM.ai
GPT-5.4 MultimodalGrounded: 73.1 · BenchLM.ai
GPT-5.4 InstructionFollowing: 90.7 · BenchLM.ai
GPT-5.4 Humanity's Last Exam: 44.3 · Epoch AI
GPT-5.4 LM Arena: 1475.121670399431 · LM Arena
GPT-5.5 Agentic: 62.71 · BenchLM.ai
GPT-5.5 GDP.pdf: 26 · Epoch AI
GPT-5.5 LM Arena: 1481.4402312219954 · LM Arena
GPT-5.5 MultimodalGrounded: 75.5 · BenchLM.ai
GPT-5.5 Coding: 62.06 · BenchLM.ai
GPT-5.5 Knowledge: 72.04 · BenchLM.ai
GPT-5.5 InstructionFollowing: 93.3 · BenchLM.ai
GPT-5.5 ExploitBench: 41.8 · Epoch AI
GPT-5.5 Reasoning: 70.4 · BenchLM.ai
GPT-5.5 Instant LM Arena: 1473.1071293051452 · LM Arena
GPT-5.6 Luna Coding: 61.65 · BenchLM.ai
GPT-5.6 Luna Agentic: 55.96 · BenchLM.ai
GPT-5.6 Luna Knowledge: 64.76 · BenchLM.ai
GPT-5.6 Luna MultimodalGrounded: 79.6 · BenchLM.ai
GPT-5.6 Luna Reasoning: 58.2 · BenchLM.ai
GPT-5.6 Luna LM Arena: 1452.5662458726235 · LM Arena
GPT-5.6 Sol Coding: 67.47 · BenchLM.ai
GPT-5.6 Sol Agentic: 68.65 · BenchLM.ai
GPT-5.6 Sol Knowledge: 77.3 · BenchLM.ai
GPT-5.6 Sol MultimodalGrounded: 89.8 · BenchLM.ai
GPT-5.6 Sol InstructionFollowing: 89.1 · BenchLM.ai
GPT-5.6 Sol GDP.pdf: 30.7 · Epoch AI
GPT-5.6 Sol Reasoning: 76.5 · BenchLM.ai
GPT-5.6 Sol LM Arena: 1484.7394950911005 · LM Arena
GPT-5.6 Terra Coding: 61.97 · BenchLM.ai
GPT-5.6 Terra Agentic: 60.37 · BenchLM.ai
GPT-5.6 Terra Knowledge: 70.2 · BenchLM.ai
GPT-5.6 Terra MultimodalGrounded: 84.5 · BenchLM.ai
GPT-5.6 Terra InstructionFollowing: 87.1 · BenchLM.ai
GPT-5.6 Terra Reasoning: 69.1 · BenchLM.ai
GPT-5.6 Terra LM Arena: 1466.429395420974 · LM Arena
GPT-6 Astra Coding: 75.56 · BenchLM.ai
GPT-6 Astra Agentic: 75.09 · BenchLM.ai
GPT-6 Astra Reasoning: 93.3 · BenchLM.ai
GPT-6 Astra Knowledge: 86.24 · BenchLM.ai
GPT-6 Astra Humanity's Last Exam: 54.8 · Epoch AI
GPT-6 Astra LM Arena: 1477.319231586918 · LM Arena
GPT-6 Luna LM Arena: 1442.8129943072395 · LM Arena
GPT-6 Luna Coding: 53.3 · BenchLM.ai
GPT-6 Luna Agentic: 56.44 · BenchLM.ai
GPT-6 Luna Reasoning: 59.6 · BenchLM.ai
GPT-6 Luna Knowledge: 67.44 · BenchLM.ai
GPT-6 Sol Coding: 66.41 · BenchLM.ai
GPT-6 Sol Agentic: 58.16 · BenchLM.ai
GPT-6 Sol Reasoning: 73.1 · BenchLM.ai
GPT-6 Sol Knowledge: 80.59 · BenchLM.ai
GPT-6 Sol LM Arena: 1459.0243436054134 · LM Arena
GPT-6.1 Sol Coding: 71.04 · BenchLM.ai
GPT-6.1 Sol Agentic: 76.22 · BenchLM.ai
GPT-6.1 Sol Reasoning: 90.5 · BenchLM.ai
GPT-6.1 Sol Knowledge: 86.09 · BenchLM.ai
GPT-6.1 Sol LM Arena: 1483.8469422064072 · LM Arena
gpt-oss SWE-Bench Pro: 16.2 · Hugging Faceself reported
gpt-oss LM Arena: 1351.758607712953 · LM Arena
gpt-oss InstructionFollowing: 84.3 · BenchLM.ai
Granite 3 LM Arena: 1183.260094120219 · LM Arena
Granite 4 LM Arena: 1287.4742107786171 · LM Arena
Granite 4.1 LM Arena: 1304.1677821649496 · LM Arena
Grok 4 LM Arena: 1410.8508236600535 · LM Arena
Grok 4.1 LM Arena: 1459.0230228462426 · LM Arena
Grok 4.20 MultimodalGrounded: 53.4 · BenchLM.ai
Grok 4.3 GDP.pdf: 8 · Epoch AI
Grok 4.3 LM Arena: 1441.7179508681295 · LM Arena
Grok 4.3 InstructionFollowing: 97.1 · BenchLM.ai
Grok 4.3 MultimodalGrounded: 78.8 · BenchLM.ai
Grok 4.5 LM Arena: 1465.7091491137166 · LM Arena
Grok 4.5 Coding: 55.72 · BenchLM.ai
Grok 4.5 Agentic: 58.85 · BenchLM.ai
Grok 4.5 Knowledge: 67.43 · BenchLM.ai
Grok 4.5 Reasoning: 54.1 · BenchLM.ai
Grok 4.6 Coding: 58.41 · BenchLM.ai
Grok 4.6 Agentic: 69.73 · BenchLM.ai
Grok 4.6 Knowledge: 69.2 · BenchLM.ai
Grok 4.6 Reasoning: 61.4 · BenchLM.ai
Grok 4.6 LM Arena: 1454.4891469812887 · LM Arena
Grok 4.7 Reasoning: 59.8 · BenchLM.ai
Grok 4.7 LM Arena: 1443.9789336394695 · LM Arena
Kimi K2 SWE-Bench Pro: 27.7 · Hugging Faceself reported
Kimi K2.5 MultimodalGrounded: 70.4 · BenchLM.ai
Kimi K2.5 Multilingual: 87.3 · BenchLM.ai
Kimi K2.5 InstructionFollowing: 85.8 · BenchLM.ai
Kimi K2.5 Humanity's Last Exam: 24.4 · Epoch AI
Kimi K2.5 SWE-Bench Pro: 50.7 · Hugging Faceself reported
Kimi K2.6 MultimodalGrounded: 73.8 · BenchLM.ai
Kimi K2.6 GDP.pdf: 12 · Epoch AI
Kimi K2.6 ExploitBench: 16.3 · Epoch AI
Kimi K2.6 SWE-Bench Pro: 58.6 · Hugging Faceself reported
Kimi K2.6 LM Arena: 1460.8290167351456 · LM Arena
Kimi K2.6 Math: 76.8 · BenchLM.ai
Kimi K2.6 Knowledge: 60.3 · BenchLM.ai
Kimi K3 LM Arena: 1487.071983223727 · LM Arena
Kimi K3 MultimodalGrounded: 92.6 · BenchLM.ai
Kimi K3 Coding: 58.91 · BenchLM.ai
Kimi K3 Agentic: 69.34 · BenchLM.ai
Kimi K3 Knowledge: 68.74 · BenchLM.ai
Kimi K3 Reasoning: 70 · BenchLM.ai
Llama 3.3 LM Arena: 1318.0654473917496 · LM Arena
Llama 4 Humanity's Last Exam: 5.7 · Epoch AI
Llama 4 LM Arena: 1326.9823064447999 · LM Arena
MiniMax M2 LM Arena: 1343.3349893374027 · LM Arena
MiniMax M2.1 SWE-Bench Pro: 36.8 · Hugging Faceself reported
MiniMax M2.1 LM Arena: 1384.437854918656 · LM Arena
MiniMax M2.5 SWE-Bench Pro: 55.4 · Hugging Faceself reported
MiniMax M2.5 LM Arena: 1390.6393248159263 · LM Arena
MiniMax M2.7 InstructionFollowing: 93 · BenchLM.ai
MiniMax M2.7 ExploitBench: 13.3 · Epoch AI
MiniMax M2.7 SWE-Bench Pro: 56.2 · Hugging Faceself reported
MiniMax M2.7 LM Arena: 1415.0035341569967 · LM Arena
MiniMax M3 SWE-Bench Pro: 59 · Hugging Faceself reported
MiniMax M3 LM Arena: 1441.3155955735351 · LM Arena
MiniMax-M1 LM Arena: 1363.8032373271456 · LM Arena
Mistral 7B LM Arena: 1110.4879069188864 · LM Arena
Mistral Large LM Arena: 1242.6627869860827 · LM Arena
Mistral Large 2 LM Arena: 1305.5745826443517 · LM Arena
Mistral Large 3 GDP.pdf: 2 · Epoch AI
Mistral Large 3 LM Arena: 1414.2471003865203 · LM Arena
Mistral Large 4 Agentic: 50.16 · BenchLM.ai
Mistral Large 4 Reasoning: 80.8 · BenchLM.ai
Mistral Large 4 LM Arena: 1431.770090901766 · LM Arena
Mistral Medium 3 Humanity's Last Exam: 4.5 · Epoch AI
Mixtral 8x22B LM Arena: 1229.7102453695306 · LM Arena
Mixtral 8x7B LM Arena: 1197.2836860825435 · LM Arena
Muse Glimmer 30B LM Arena: 1424.5427613711804 · LM Arena
Muse Glimmer 30B MultimodalGrounded: 65.1 · BenchLM.ai
Muse Glimmer 30B SWE-Bench Pro: 51.2 · Hugging Faceself reported
Muse Spark Humanity's Last Exam: 40.6 · Epoch AI
Muse Spark LM Arena: 1489.0413932749987 · LM Arena
Muse Spark Coding: 49.95 · BenchLM.ai
Muse Spark Agentic: 51.46 · BenchLM.ai
Muse Spark MultimodalGrounded: 84.4 · BenchLM.ai
Muse Spark Knowledge: 62.5 · BenchLM.ai
Muse Spark InstructionFollowing: 93.3 · BenchLM.ai
Muse Spark Open Coding: 56.11 · BenchLM.ai
Muse Spark Open Agentic: 60.11 · BenchLM.ai
Muse Spark Open Knowledge: 64.69 · BenchLM.ai
Nemotron 3 Multilingual: 89.2 · BenchLM.ai
Nemotron 3 InstructionFollowing: 92.1 · BenchLM.ai
Nemotron 3 GDP.pdf: 2 · Epoch AI
Nemotron-4 340B LM Arena: 1277.2503679791023 · LM Arena
North Mini Code SWE-Bench Pro: 40.2 · Hugging Faceself reported
o1 InstructionFollowing: 86 · BenchLM.ai
o1 Humanity's Last Exam: 8.1 · Epoch AI
o1 LM Arena: 1402.2836309363597 · LM Arena
o3 Humanity's Last Exam: 20.3 · Epoch AI
o3 LM Arena: 1431.6911628578932 · LM Arena
o3 InstructionFollowing: 87.4 · BenchLM.ai
o3-mini LM Arena: 1363.5818572513192 · LM Arena
o4-mini Humanity's Last Exam: 18.1 · Epoch AI
o4-mini LM Arena: 1391.0547755412438 · LM Arena
Phi-3 LM Arena: 1128.2044719870635 · LM Arena
Phi-4 LM Arena: 1256.3106977286461 · LM Arena
Phi-4 Multilingual: 1 · BenchLM.ai
Qwen 2.5 LM Arena: 1373.7691794399948 · LM Arena
Qwen 3 SWE-Bench Pro: 21.4 · Hugging Faceself reported
Qwen 3 LM Arena: 1422.131004335956 · LM Arena
Qwen 3 Multilingual: 83.8 · BenchLM.ai
Qwen 3.5 MultimodalGrounded: 73.6 · BenchLM.ai
Qwen 3.5 Multilingual: 93.8 · BenchLM.ai
Qwen 3.5 InstructionFollowing: 93 · BenchLM.ai
Qwen 3.5 LM Arena: 1441.5655970064536 · LM Arena
Qwen 3.6 MultimodalGrounded: 76.1 · BenchLM.ai
Qwen 3.6 Multilingual: 93.8 · BenchLM.ai
Qwen 3.6 InstructionFollowing: 85.8 · BenchLM.ai
Qwen 3.6 SWE-Bench Pro: 53.5 · Hugging Faceself reported
Qwen 3.6 Math: 69.5 · BenchLM.ai
Qwen 3.7 Max Knowledge: 61.33 · BenchLM.ai
Qwen 3.7 Max Multilingual: 100 · BenchLM.ai
Qwen 3.7 Max InstructionFollowing: 92.8 · BenchLM.ai
Qwen 3.7 Max LM Arena: 1474.346453572102 · LM Arena
Qwen 3.7 Max Math: 79 · BenchLM.ai
Qwen3-Coder SWE-Bench Pro: 38.7 · Hugging Faceself reported
Qwen3-Coder LM Arena: 1387.5520617800003 · LM Arena
Qwen3-Coder-Next SWE-Bench Pro: 44.3 · Hugging Faceself reported
Qwen3-Max LM Arena: 1434.162064836811 · LM Arena
QwQ-32B LM Arena: 1335.575784131921 · LM Arena

News & rumors added

NewsGPT-5.6 Sol

OpenAI publishes coding evaluations for low-resource languages

OpenAI introduced MultiLangBench, which measures whether models can build SQLite-compatible database engines across 20 languages, and a separate evaluation of COBOL migrations.…

OpenAI · Oct 19, 2026

Read at openai.com

Every story we’re tracking →