OpenAI's GPT-5.4 mini now available in GitHub Copilot
OpenAI has released GPT-5.4 mini, the lightweight variant of its agentic coding model GPT-5.4, in GitHub Copilot. The model represents OpenAI's highest-performing mini offering to date for code generation and completion tasks.
GPT-5.4 mini — Quick Specs
OpenAI's GPT-5.4 mini Now Generally Available in GitHub Copilot
OpenAI's GPT-5.4 mini has begun rolling out to GitHub Copilot users as a generally available option. According to GitHub, the model is the latest fast-optimized version of GPT-5.4, OpenAI's agentic coding model designed for code generation and completion.
Performance Claims
GitHub states that GPT-5.4 mini represents OpenAI's highest-performing mini model to date in early internal testing. The company has not yet disclosed specific benchmark scores, latency metrics, or token pricing for the model variant.
What's Included
GPT-5.4 mini joins GitHub Copilot's existing model roster, which already includes access to various OpenAI models. The release appears focused on providing developers with a faster, more cost-efficient option for real-time code completion while maintaining competitive performance compared to previous mini variants.
No specific context window size, input/output pricing per million tokens, or parameter count has been disclosed in the announcement.
What This Means
GPT-5.4 mini positions GitHub Copilot users to access a presumably faster inference speed relative to full GPT-5.4, which could reduce latency in IDE integration. However, without published benchmarks or pricing, the concrete advantages over existing models remain unclear. The release follows the industry pattern of offering tiered model variants—full-scale and mini editions—to balance performance with speed and cost across different use cases. Developers using GitHub Copilot should expect this as an additional option within their existing subscription tier.
Related Articles
Robot Safety Benchmark Finds GPT-6 Astra and Claude Fable 5.1 Rarely Refuse Dangerous Commands
A new benchmark called RoboHarm tested whether AI models controlling robotic arms would refuse dangerous commands. GPT-6 Astra completed 60 of 100 dangerous tasks and Claude Fable 5.1 completed 34, with neither model showing a reliable safety layer.
Xiaomi Releases MiMo-V2.6-Pro-RL, a 1.02T-Parameter Omnimodal Model with 1M-Token Context
Xiaomi's MiMo team has released MiMo-V2.6-Pro-RL, a 1.02-trillion-parameter sparse mixture-of-experts model with 42B active parameters, 1M-token context, and native text/image/video/audio processing. The model was trained via a single mixed reinforcement learning run spanning coding, agentic, visual, and cybersecurity tasks, with benchmark scores that Xiaomi claims approach or match Claude Opus 5 and GPT-5.6 on several agentic and coding tests.
Xiaomi Releases MiMo-V2.6-Flash-RL, a 309B-Parameter MoE Model with 1M-Token Context and Native Omnimodal Support
Xiaomi's MiMo team released MiMo-V2.6-Flash-RL, an efficiency-tier checkpoint in the MiMo-V2.6 series featuring a 309B-parameter (15B active) Mixture-of-Experts architecture, 1M-token context, and native support for text, image, video, and audio. The model uses a single mixed reinforcement learning run across coding, agentic, visual, and cybersecurity tasks rather than domain-specific training.
TypeSafe AI Launches Jev, a 'Decision Model' That Outputs Only Numbers, Priced at $0.042/M Input Tokens
TypeSafe AI has released Jev, the first model in a new category it calls 'System One models'—text goes in, floating-point decisions come out. At $0.042 per million input tokens with free output, it undercuts even GPT-5 Nano on price.
Comments
Loading...