Who contributed to Google on August 5, 2026?

8 developers shipped this update, including copybara-service, jakevdp, ashahba, Annie Luc, shuoweil, daniel-sanche, Dev Variya, and dependabot.

What were the notable Google updates?

[Mosaic TPU] Support lane dimension argmax/argmin with bf16 input, Add jax.numpy.top_k, and [JAX:GPU][oneAPI]: Add OneAPI CI test scripts and infrastructure.

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 ADDS NUMPY 2.6 TOP_K, INTEL GPU SUPPORT LANDS IN CI

By RepoJournal · Filed 06:59 UTC on August 5, 2026 · About Google

8 people shipped this

copybara-service @copybara-service 2 cited

jakevdp @jakevdp 1 cited

ashahba @ashahba 1 cited

shuoweil @shuoweil 1 cited

daniel-sanche @daniel-sanche 1 cited

Dev Variya @fixedbydev 1 cited

Annie Luc @Annhiluc

dependabot @dependabot

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.

JAX landed two major capability expansions this cycle. First, `jax.numpy.top_k` now implements parity with NumPy v2.6.0 [2], tested against both the NumPy nightly and fallback baselines to ensure compatibility across versions. Simultaneously, OneAPI CI infrastructure hit the JAX tree [3], adding automated GPU detection, parallelization scripts, and the full test harness for Intel GPU backends. This represents the completion of GPU vendor diversity in JAX's continuous integration pipeline.

On the performance front, JAX added native support for lane dimension argmax/argmin operations with bf16 inputs on Mosaic TPU [1], enabling efficient 16-bit instruction paths via `lax.argmax/argmin` with `index_type=int16`. The team also fixed a reduce_max/reduce_min operand passthrough issue and added upstream rules for div [4] to resolve JVP call chains.

Down the stack, google-cloud-python addressed critical operational issues. A debug print statement was leaking to stdout in every BigQuery RangeQueryParameter construction [7], now removed. The team also standardized x-goog-api-client header handling to prevent duplication across the auth/core/user layers [6], reducing backend compatibility friction. Pandas-GBQ version detection gained a capability helper for safe delegation and telemetry [5], handling missing or corrupted installations gracefully.

Action items

→ Test JAX workloads against top_k for NumPy 2.6 compatibility google/jax [plan]
→ Upgrade google-cloud-bigquery immediately if using RangeQueryParameter to stop debug output googleapis/google-cloud-python [immediate]
→ Review x-goog-api-client header deduplication in your GCP client calls googleapis/google-cloud-python [monitor]

References

[1] [Mosaic TPU] Support lane dimension argmax/argmin with bf16 input. ↗ google/jax
[2] Add jax.numpy.top_k ↗ google/jax
[3] [JAX:GPU][oneAPI]: Add OneAPI CI test scripts and infrastructure ↗ google/jax
[4] Allow `reduced` operand to pass through `reduce_{max,min}`. Also add ur_rule for `div_p` since `reduce_max` jvp calls into div. ↗ google/jax
[5] feat: add pandas-gbq capability helper ↗ googleapis/google-cloud-python
[6] fix: deduplicate x-goog-api-client headers ↗ googleapis/google-cloud-python
[7] fix: remove stray debug print in RangeQueryParameter constructor (#17973) googleapis/google-cloud-python

Quick answers

What shipped in Google on August 5, 2026?: 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. In total, 27 commits and 32 pull requests landed.
Who contributed to Google on August 5, 2026?: 8 developers shipped this update, including copybara-service, jakevdp, ashahba, Annie Luc, shuoweil, daniel-sanche, Dev Variya, and dependabot.
What were the notable Google updates?: [Mosaic TPU] Support lane dimension argmax/argmin with bf16 input, Add jax.numpy.top_k, and [JAX:GPU][oneAPI]: Add OneAPI CI test scripts and infrastructure.

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 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

Google Aug 2

JAX AVERTS BREAKING CHANGE IN CUSTOM_GRADIENT API

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

ai 3 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.