aprender 0.70.3

aprender 0.70.3 — release notes

0.70.3 is a patch release that carries one change: streaming tool calls (Refs #4918).
Its published-path diff against v0.70.2 (crates, src, Cargo.toml, Cargo.lock) is the #4918
fix's files plus the version bump, and nothing else.

How this release was gated

0.70.3 ships under a recorded operator emergency scope, crux-smoke, as 0.70.2, 0.70.1 and 0.69.1 did.
The operator's words: "0.70.3 ships on CRUX smoke on lambda and gx10 GPU. Everything bigger is nightly."
The record is the first emergency_scopes entry, crux-smoke for release "0.70.3", in
contracts/model-capability-ladder-v1.yaml. The standing policy there still starts at 0.71.0.

The gate. The CRUX smoke runs on the RTX 4090 and the GB10, with nothing else on either GPU, over every
certified model, both thinking modes, the control prompts, and the verbs run, chat, serve and code, with the
binary built at the release commit. Those receipts are published with the GitHub release for v0.70.3.

The prompt certification is carried from 0.70.2, not re-run.
evidence/crux/0.70.3/prompt-certification.json and prompt-certification-inventory.json are byte copies of
the evidence/crux/0.70.2/ files. Between v0.70.2 and this release's base, evidence/crux/0.70.1, the 0.69.1
prompt certification, scripts/crux_inference_dogfood.sh, scripts/lib/crux_smoke_scope.py and
scripts/crux_sweep_shards.sh have no diff, and the certified model files are the ones pinned by sha256
inside the certification.

Unchanged. Clean tree, tag on HEAD, release branch, dogfood GO for this commit, and the clean-room green on
the tagged commit.

Changed since 0.70.2

• apr serve: with tools in a request, a streamed chat (stream: true on /v1/chat/completions, and
  /v1/chat/completions/stream) now sends the model's tool call as delta.tool_calls and ends with
  finish_reason: "tool_calls", as the non-streaming path already did. Both paths parse the call with the
  same parser. With no tools, or tool_choice: "none", the stream is unchanged (Refs #4918).

Known failures (shipped, moved to 0.71.0)

The operator refused all six for 0.70.2 and they stand for 0.70.3 under the scope above. None has a per-model contract edit, and none is
marked known red or green. Each stays red in the printed evidence and in its ticket, and each is on the 0.71.0
milestone. Each was re-measured on the RTX 4090 on 2026-10-06, against 0.70.1 built from main, and still fails the same way.

model	GPU	check	failure	ticket
qwen35-0.8b-q4km	RTX 4090, GB10	golden_output thinking_on	think block unclosed within 2048 tokens	#4666
Qwen2.5-0.5B-Instruct-f16.gguf	RTX 4090	cuda serve /api/chat	gibberish output	#4661
Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf	RTX 4090, GB10	cuda serve /api/chat	gibberish output	#4662
Qwen3.5-0.8B-IQ4_XS.gguf	RTX 4090	golden_output thinking_on	no measured thinking budget (named refusal)	#4663
Qwen3.5-0.8B-UD-IQ2_XXS.gguf	RTX 4090, GB10	golden_output	GPU forward fell back to CPU (run and chat rc=14); apr qa also fails tensor_contract	#4664
Qwen3.5-35B-A3B-UD-IQ4_XS.gguf	RTX 4090	capability_match	no CUDA forward for qwen35moe (named refusal)	#4665

Not measured for 0.70.3

• Capability cells (model_ladder.sh --cells), as for 0.70.2: a gap, not a pass
  (#4667).
• The full CRUX sweep (beyond the smoke above) and CPU lanes: nightly, per the scope.