OpenAI Model Tops RISC-V Core Tests

GPT-6.2 Astra posted the highest average score across two open-source RISC-V processor-core tests, while OpenAI models improved GPU IP metrics.

Updated on Nov. 19, 2026 in Artificial Intelligence

Close-up photograph of a bare silicon wafer with a single square processor die in sharp focus under cool directional light on a dark matte surface.
OpenAI's GPT-6.2 Astra posted the highest average score on two open-source RISC-V processor-core tests, and its models improved GPU IP metrics by up to 14%. AI Illustration. Upload story photo >

Live Poll

Do you think AI is making everyday technology better for you?

OpenAI's GPT-6.2 Astra recorded the highest average score across two open-source RISC-V processor-core tests. OpenAI models also produced up to a 14% improvement in power, performance, and area on flagship GPU IP.

Why it matters

The results show AI models being applied to hardware engineering, not only software tasks, which could shorten chip design cycles. The work also points toward AI helping optimize the very chips that run AI workloads.

The evaluations used two open-source RISC-V processor-core tests, with each run budgeted at about 112,000 tokens. OpenAI models delivered up to 14% power, performance, and area gains on flagship GPU IP, though the baseline comparisons were not detailed.

The players

OpenAI

OpenAI is an American artificial intelligence research company known for developing the GPT family of large language models and ChatGPT.

The details

An OpenAI model helped optimize Jalapeño, the company's first custom inference chip, and those improvements were validated on a physical chip. The team is also working to improve the model's communication style after receiving feedback.

Timeline

  1. Update describing the RISC-V results and chip-design work was published on November 19, 2026.

The Tech Race

The work places OpenAI's Jalapeño inference chip within a broader effort by AI developers to build custom silicon rather than rely solely on general-purpose GPUs. It also signals that AI-assisted chip design is moving from theory into validated hardware results.

Faster, AI-assisted chip design could eventually lead to more efficient processors in consumer devices and data centers. Users are unlikely to see direct changes soon, since these results come from internal evaluations and early hardware work.

The takeaway

AI models are beginning to contribute to hardware engineering, and validated gains on physical chips suggest the approach can move beyond benchmarks. Readers following AI infrastructure should watch whether these optimizations reach production products.

Further reading

Explore more Artificial Intelligence coverage for context on how AI models are reshaping hardware and software development.

Source note: This article includes information reported by TokenPost.

Live Poll

Do you think AI is making everyday technology better for you?