Who contributed to Google on August 1, 2026?

8 developers shipped this update, including mattjj, mraunak, copybara-service, Google Team Member, Matthew Tang, hebaalazzeh, chalmerlowe, and sakthivelmanii.

What were the notable Google updates?

[vjp3] add defvjp_with_logs so custom_vjp rules can log, remove qdd, and remove os.subprocess test.

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 REFACTORS CORE VJP SYSTEM WHILE CLOUD PYTHON FIXES SPANNER JSON BUGS

By RepoJournal · Filed 06:59 UTC on August 1, 2026 · About Google

8 people shipped this

mattjj @mattjj 3 cited

sakthivelmanii @sakthivelmanii 2 cited

copybara-service @copybara-service 1 cited

hebaalazzeh @hebaalazzeh 1 cited

mraunak @mraunak

Google Team Member @google-genai-bot

Matthew Tang @matthew29tang

chalmerlowe @chalmerlowe

JAX's custom_vjp rules gain logging capability as the team strips deprecated code, while google-cloud-python patches critical JsonObject serialization breaks across nested structures.

JAX shipped a major VJP refactor [1] that implements defvjp_with_logs on both classic custom_vjp and the newer custom_vjp3, letting gradient rules log computation steps without breaking downstream abstractions. The team is simultaneously cleaning house, removing the deprecated QDD module [2] and subprocess test code [3] to reduce maintenance surface. On GPU acceleration, Pallas now transparently routes gmem-to-smem copies through TMA on Hopper+ hardware while falling back to cp.async on older architectures [4], a quiet win for multi-GPU workloads. Over on google-cloud-python, a critical fix landed for Spanner's JsonObject [5] that restores standard Python container protocols (len, bool, iter, getitem) on array/scalar/null variants, addressing data erasure bugs in nested serialization [6]. The cloud team also refactored generator templates to delegate REST transcoding to google-api-core [7], centralizing request handling logic that was previously scattered across three separate helper functions. Activity across the three repos: 44 commits, 36 PRs in the last 24 hours.

Action items

→ Test JAX custom_vjp3 logging in your gradient rules before next release google/jax [plan]
→ Upgrade google-cloud-python to fix Spanner JsonObject container protocol breaks immediately googleapis/google-cloud-python [immediate]
→ Review REST transcoding refactor in google-api-core delegation for API compatibility googleapis/google-cloud-python [plan]

References

[1] [vjp3] add defvjp_with_logs so custom_vjp rules can log ↗ google/jax
[2] remove qdd ↗ google/jax
[3] remove os.subprocess test google/jax
[4] [pallas:mgpu] `copy_gmem_to_smem` now transparently uses TMA on Hopper+ and cp.async on older architectures ↗ google/jax
[5] fix(spanner): implement dict protocol and nested unwrapping for JsonObject ↗ googleapis/google-cloud-python
[6] fix(spanner): implement dict protocol and nested unwrapping for JsonObject (#17915) googleapis/google-cloud-python
[7] feat(generator): delegate REST transcoding to google-api-core ↗ googleapis/google-cloud-python

Quick answers

What shipped in Google on August 1, 2026?: JAX's custom_vjp rules gain logging capability as the team strips deprecated code, while google-cloud-python patches critical JsonObject serialization breaks across nested structures. In total, 44 commits and 36 pull requests landed.
Who contributed to Google on August 1, 2026?: 8 developers shipped this update, including mattjj, mraunak, copybara-service, Google Team Member, Matthew Tang, hebaalazzeh, chalmerlowe, and sakthivelmanii.
What were the notable Google updates?: [vjp3] add defvjp_with_logs so custom_vjp rules can log, remove qdd, and remove os.subprocess test.

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.