Who contributed to Google on August 2, 2026?

3 developers shipped this update, including Yash Katariya, copybara-service, and jax authors.

What were the notable Google updates?

Don't make `fun` in custom_gradient positional-only since that's a breaking change, Don't make `fun` in custom_gradient positional-only since that's a breaking change, and Update XLA dependency to use revision http://github.com/openxla/xla/commit/4f8d2e3c11bbe311df280b33d59dd8569df6b1db.

Google

JAX, the GenAI SDK, and the Cloud libs - Google's open source layer

Pick a date

Topics: AI / ML Python Full archive →

The Wire · Showcase

JAX AVERTS BREAKING CHANGE IN CUSTOM_GRADIENT API

By RepoJournal · Filed 06:57 UTC on August 2, 2026 · About Google

3 people shipped this

Yash Katariya @yashk2810 1 cited

copybara-service @copybara-service 1 cited

jax authors @Google-ML-Automation 1 cited

JAX pulled back from making the `fun` parameter positional-only in custom_gradient, preserving backward compatibility just before release.

The custom_gradient function was about to lose flexibility in how developers pass the `fun` argument [1], a change that would have broken existing code across the ecosystem. JAX maintainers caught this in review and reverted course [2], keeping the parameter keyword-compatible. In parallel, the team updated XLA dependencies to pull in the latest compiler improvements [3], keeping the ML stack current without collision. Two commits, one PR: minimal churn, maximum fidelity. This is exactly how you handle API stability at scale.

Action items

→ No action required - JAX preserved backward compatibility google/jax [monitor]
→ Watch for official 0.x release notes when XLA integration lands google/jax [monitor]

References

[1] Don't make `fun` in custom_gradient positional-only since that's a breaking change google/jax
[2] Don't make `fun` in custom_gradient positional-only since that's a breaking change ↗ google/jax
[3] Update XLA dependency to use revision http://github.com/openxla/xla/commit/4f8d2e3c11bbe311df280b33d59dd8569df6b1db google/jax

Quick answers

What shipped in Google on August 2, 2026?: JAX pulled back from making the `fun` parameter positional-only in custom_gradient, preserving backward compatibility just before release. In total, 2 commits and 1 pull requests landed.
Who contributed to Google on August 2, 2026?: 3 developers shipped this update, including Yash Katariya, copybara-service, and jax authors.
What were the notable Google updates?: Don't make `fun` in custom_gradient positional-only since that's a breaking change, Don't make `fun` in custom_gradient positional-only since that's a breaking change, and Update XLA dependency to use revision http://github.com/openxla/xla/commit/4f8d2e3c11bbe311df280b33d59dd8569df6b1db.

JAX PINS PYTHON 3.14T, GENAI GAINS ROBOTICS MODEL AND ERROR HANDLING

JAX disabled profiler tests on free-threaded Python to unblock CI, while python-genai 2.17.0 ships the Gemini Robotics ER 2 Preview and fixes tool call limits.

ai 90 shipped

Google Aug 5

JAX ADDS NUMPY 2.6 TOP_K, INTEL GPU SUPPORT LANDS IN CI

JAX shipped native top_k alignment with NumPy 2.6.0 while simultaneously landing full OneAPI CI infrastructure for Intel GPU testing across the entire stack.

ai 59 shipped

Google Aug 4

JAX HARDENS GPU KERNELS WHILE CLOUD PYTHON PATCHES SUPPLY CHAIN RISK

JAX shipped critical Mosaic GPU optimizations that squeeze performance out of transposed memory transfers, while the cloud SDK locked down six open security vulnerabilities in sqlalchemy-spanner.

ai 100 shipped

Google Aug 3

JAX SYNCS XLA DEPENDENCY TO LATEST REVISION

JAX pinned its XLA dependency to a fresh upstream revision overnight, pulling in compiler optimizations and fixes across the ML stack.

ai 1 shipped

Elsewhere on the wire

AI Agents Aug 6

LANGCHAIN HARDENS ANTHROPIC TOOL CALLING AND GIT CREDENTIALS AS OPENHANDS SHIPS 1.10.0

LangChain fixed a critical credential leak in evaluation runs while Anthropic tool schemas get proper fallback handling, as three major agent frameworks push releases across the ecosystem.

ai-agents 54 shipped

Local LLMs Aug 6

OLLAMA FIXES KV CACHE CORRUPTION, LLAMA.CPP HARDENS FILE SEARCH, VLLM KILLS MOE LEGACY CODE

Ollama patched a critical cache corruption bug that corrupts lazy snapshots during multi-request prefills, while llama.cpp hardened Windows junction traversal and vLLM removed 14 hardcoded MoE shards across quantization methods.