P P99lab Meta OpenAI DeepSeek Head to head on 25 boards: OpenAI places best on 17 , Meta places best on 7 , P99lab places best on 1 and DeepSeek places best on 1 (a tie counts for each). All 4 ranked on 0 of 26 boards.
Board Read P P99lab Meta OpenAI DeepSeek Chat2 MultiChallenge Score 1d yesterday Oct 26 1d yesterday #3 75.5% Muse Spark #1 82.0% best of the picked GPT-6.1 Sol #28 46.1% DeepSeek V3P1 Text Arena Arena rating 1d yesterday Oct 26 1d yesterday #3 1,509 best of the picked Muse Wonder #18 1,485 GPT-5.6 Sol #29 1,476 DeepSeek V4.1 Flash Reasoning7 ARC-AGI-2 Score 1d yesterday Oct 26 1d yesterday #38 57.5% Muse Spark Open #1 95.0% best of the picked GPT-6 Astra #18 78.1% DeepSeek V4.1 Flash GPQA Diamond Accuracy 1d yesterday Oct 26 1d yesterday #43 89.8% Muse Spark #2 95.8% best of the picked GPT-6 Astra #19 93.2% DeepSeek V4.1 Flash Epoch ECI ECI 1d yesterday Oct 26 1d yesterday #12 161.8 Muse Wonder #2 168.8 best of the picked GPT-6 Astra #32 155.9 DeepSeek V4.1 Flash AA Intelligence Index 1d yesterday Oct 26 1d yesterday #3 55.8 best of the picked Muse Wonder Full #6 52.7 GPT-6 Astra #14 45.1 DeepSeek V4.1 Flash MMLU-Pro Accuracy · stale 1d yesterday Oct 26 1d yesterday #47 82.8% Llama4 Behemoth #3 89.3% best of the picked o1 #33 85.0% DeepSeek-V3.2 LiveBench Global average 1d yesterday Oct 26 1d yesterday #6 81.6% Muse Spark 1.3 #4 82.2% best of the picked GPT-6 Astra #8 81.1% DeepSeek V4.1 Flash HLE Accuracy 1d yesterday Oct 26 1d yesterday #7 40.6% Muse Spark #1 54.8% best of the picked GPT-6 Astra Coding4 Aider Polyglot Pass rate · stale 1d yesterday Oct 26 1d yesterday #43 15.6% Llama 4 Maverick Instruct #1 88.0% best of the picked GPT-5 #8 74.2% DeepSeek-V3.2 WebDev Arena Arena rating 1d yesterday Oct 26 1d yesterday #6 1,726 Muse Wonder Full #2 1,774 best of the picked GPT-6 Astra #12 1,665 DeepSeek V4.1 Flash SWE-bench Verified Resolved · stale 1d yesterday Oct 26 1d yesterday #36 21.0% Llama 4 Maverick Instruct #6 72.8% best of the picked GPT-5.2 #11 70.0% DeepSeek-V3.2 SWE-Bench Pro Resolved 1d yesterday Oct 26 1d yesterday #4 90.2% best of the picked Muse Wonder 1.0 Max Effort #4 90.2% best of the picked GPT-6 Astra Agents and tools3 BFCL Overall accuracy · stale 1d yesterday Oct 26 1d yesterday #36 37.3% Llama 4 Maverick 17b 128e #7 63.0% best of the picked o3 #12 56.7% DeepSeek-V3.2 MCP Atlas Pass rate 1d yesterday Oct 26 1d yesterday #1 88.1% best of the picked Muse Spark 1.1 #11 81.8% GPT-5.6 Search Arena Arena rating 1d yesterday Oct 26 1d yesterday #1 1,258 GPT-5.6 Sol Vision2 Vision Arena Arena rating 1d yesterday Oct 26 1d yesterday #6 1,294 best of the picked Muse Spark #7 1,294 GPT-6.1 Sol #39 1,261 DeepSeek V4.1 Flash Document Arena Arena rating 1d yesterday Oct 26 1d yesterday #9 1,479 Muse Spark 1.1 #6 1,489 best of the picked GPT-5.6 Sol Image2 Text-to-Image Arena rating 1d yesterday Oct 26 1d yesterday #8 1,274 Muse Image #1 1,425 best of the picked GPT Image 2.5 Sunburst Image Edit Arena rating 1d yesterday Oct 26 1d yesterday #9 1,398 Muse Image #1 1,514 best of the picked GPT Image 2.5 Sunburst Video1 Text-to-Video Arena rating 1d yesterday Oct 26 1d yesterday #9 1,456 best of the picked Muse Video #11 1,368 Sora 2 Pro Speech2 Open ASR Average WER 1d yesterday Oct 26 1d yesterday #3 4.98 WER best of the picked audrey-0.7 (p99lab) #32 6.30 WER muse-voice-transcribe (meta) #21 5.96 WER gpt-transcribe (openai) Speech Arena Arena rating 1d yesterday Oct 26 1d yesterday #95 846 Metavoice V1 #39 1,105 best of the picked TTS 1 Hd Speed1 AA Speed Output speed 1d yesterday Oct 26 1d yesterday #6 195 tok/s best of the picked Muse Wonder #12 129 tok/s GPT-6 Luna #8 181 tok/s DeepSeek V4 Flash Vision Price1 AA Price Blended price 1d yesterday Oct 26 1d yesterday #17 $2.00 Muse Spark Open #3 $0.20 best of the picked GPT-6 Luna #9 $0.53 DeepSeek V4.1 Flash Usage1 OpenRouter Weekly tokens 1d yesterday Oct 26 1d yesterday #15 2.8T tok Muse Spark 1.3 Contributor #5 8.4T tok GPT-6 Luna #1 33.6T tok best of the picked DeepSeek V4.1 Flash
Green marks the best placing among the picked where two or more are ranked; ties are marked for each. Ranks are each board's own order (lower is better), scores as the board published them. Dash: not on that board.