Microsoft releases Decision-1 AI model, claiming 83.5% JevBench score and 85ms latency
Microsoft released Microsoft-Decision-1, a decision model post-trained on Alibaba's Qwen3.5-9B, on October 12. Microsoft claims it scored 83.5% across 36 JevBench benchmarks, surpassing Quyet-1.0-Large's 81.9%, with 85-millisecond median latency and 4.5x speed over the runner-up. The model is available on OpenRouter and Microsoft Foundry, priced at $0.042 per million input tokens with free output. Microsoft says it is testing the model internally for incident response, quality control and scientific discovery.
Editorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary covers the current reports
Cross-source coverage
Reporting timeline
Microsoft Launches Decision Model Built on Alibaba's Qwen
On October 12, Microsoft released Microsoft-Decision-1, a new decision model post-trained on Alibaba's Qwen3.5-9B. Microsoft stated the model outperforms LLMs and popular decision models like Jev 1.13.0 in latency and accuracy for structured decision tasks. Microsoft's Xbox research team has used it to process over 10,000 player feedback items. CEO Satya Nadella said on social media that due to the model's strong performance, Microsoft is testing it across all areas, including incident response, quality control, and scientific discovery. The report notes that a wave of decision models, which output predictions, probabilities, and confidence scores rather than natural language, was sparked by the Jev model's popularity in September. Statistics show that of 129 open-source decision model projects on HuggingFace, 84 (over 65%) are based on Alibaba's Qwen, including models from Microsoft, Perplexity, Cloudflare, and H2O.ai.
Microsoft just shipped an AI that makes decisions faster than you can blink. A blink takes about 100 milliseconds. This one takes 85. And Satya didn't pick this day by accident. A few hours before this
Microsoft released a new AI decision model, Microsoft-Decision-1, that returns answers in 85 milliseconds, faster than a human blink at roughly 100 milliseconds. The post claims the model scores 92.2 on calibration, where 100 means confidence perfectly matches accuracy. The timing follows TypeSafe, a startup publicly active for 24 days, closing an $870 million Series A at a $7.5 billion valuation, up 37x from its mid-September seed round at $200 million. TypeSafe launched its Jev decision model on September 15, and by October 1 OpenAI, Cloudflare, AWS, and Perplexity had shipped competing decision-model offerings. Jev charges $0.042 per million input tokens with free output, while frontier LLMs take about 3 seconds on tasks the new Microsoft model clears in 85 milliseconds.
Microsoft launches Decision-1 AI model on OpenRouter, claiming top accuracy and speed
Microsoft has released its Decision-1 AI model on the OpenRouter platform. According to Microsoft's published numbers, Decision-1 achieves the highest accuracy across 36 blind benchmarks comprising approximately 150,000 questions. The model is claimed to be 4.5 times faster than the runner-up and 35 times faster than GPT-6 Sol. Additionally, its decisions flip on only 1.3% of perturbed inputs, indicating high stability. The model is post-trained from the Qwen3.5-9B base. Pricing is set at $0.042 per million input tokens, with output tokens free, and it supports a 32,000-token context window. These claims are attributed to Microsoft and have not been independently verified.
Read sourceShow 2 older updatesHide older updates
Microsoft releases Decision-1 AI model on Foundry, claims top accuracy and faster speed
Microsoft has announced the release of Microsoft-Decision-1, a new AI model for fast decision-making, now available on Microsoft Foundry. The model was post-trained on Qwen3.5-9B and is designed for single-pass decision scoring. Microsoft claims it achieved the highest accuracy in a 36-benchmark comparison spanning nearly 150,000 questions, with benchmarks kept blind from training. The company also states the model is 4.5 times quicker than Quyet-1.0-Large and 35 times quicker than GPT-6 Sol. Microsoft plans to later rebase the model on other models, including MAI and OpenAI models. The company reports it is already testing the model internally across various applications, including incident response, quality control, and scientific discovery.
Read sourceMicrosoft launches Decision-1 AI model, beating Quyet-1.0-Large on JevBench with 83.5% score
Microsoft CEO Satya Nadella announced the release of Microsoft-Decision-1, a specialized AI model designed for fast decision-making. According to the announcement, the model achieved a score of 83.5% across 36 benchmarks on JevBench, surpassing the previous leader Quyet-1.0-Large which scored 81.9%. The company claims the model is also 4.5 times faster, with a median latency of just 85 milliseconds. Microsoft is currently testing the model internally for applications in incident response, quality control, and scientific research. The announcement positions Microsoft as a direct competitor to Jev in the AI decision-making model space.