The AI Toolchain — issue -004, August 15, 2026

Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

Ollama v0.32.14 adds WebP image support and more flexible system message placement for Qwen models.

$ git clone --branch v0.32.14 https://github.com/ollama/ollama.git
# already have the repo? check out this version:
$ git checkout v0.32.14

• ›Adds WebP image transcoding for llama-server, enabling WebP inputs to be used in multimodal prompts.
• ›Qwen renderer now tolerates system messages in non-leading positions, allowing more flexible conversation structures.

Adds provider-neutral testing utilities across agents.testing, agents.realtime.testing, and agents.voice.testing, plus OpenAI Python v3 compatibility.

$ git clone --branch v0.21.0 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.21.0

• ›Adds agents.testing, agents.realtime.testing, and agents.voice.testing modules with scripted/deterministic test utilities for Agent, Sandbox, Realtime, and Voice workflows — no live provider requests required.
• ›Updates OpenAI provider compatibility to openai>=3.0.0,<4, including HTTPX2-aware request, response, transport, and exception handling.
• ›Adds configurable retry backoff ceiling for MCP connections.
• ›Adds managed_secrets support for referencing existing Runloop secrets in Sandbox sessions.

PydanticAI v2.31.0 lets UIEventStream initialize without a run_input and gives AGUIEventStream its own thread_id/run_id.

$ git clone --branch v2.31.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v2.31.0

• ›Adds support for building a UIEventStream without a run_input, enabling stream construction before run context is available.
• ›Gives AGUIEventStream its own thread_id and run_id fields for independent stream identification.

llama.cpp b10448 adds Kimi-K3 model support with hybrid KDA+MLA attention, latent MoE, and a dedicated chat format.

$ git clone --branch b10448 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b10448

• ›Adds common_chat_params_init_kimi_k3 (PEG_NATIVE) chat format for Kimi-K3 with reasoning extraction, response unwrapping, and typed tool-call parsing using the tools/call/argument tag structure.
• ›Adds kda_gate_lower_bound GGUF key (emitted by the model saver) so save/load roundtrips correctly preserve Kimi-K3's KDA gate activation; K3 sets this to -5.0.
• ›Increases LLAMA_MAX_EXPERTS from 512 to 1024 to accommodate Kimi-K3's expert count.
• ›Adds Kimi-K3 text model support with hybrid KDA (linear) + MLA (full) attention, cross-layer residual attention (attn_res_block_size), latent MoE (n_expert_latent), situ activation, MLA output gate, and full-rank KDA gate (ssm_g).
• ›Adds lossless MXFP4 repack for Kimi-K3's compressed-tensors 'mxfp4-pack-quantized' routed experts — bit-compatible with ggml MXFP4, repacked rather than dequantized, avoiding a ~5.5 TB bf16 round-trip; the repack_mxfp4_blocks helper is now shared with DeepSeek-V4.

$ git clone --branch b10444 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b10444

$ git clone --branch b10441 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b10441

$ git clone --branch b10437 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b10437

vMLX - JANGTQ Uber Compressed MLX Models - L2 Disk Cache (survives restart) + L1 Paged (super fast ttft) + Hybrid SSM Scheduler + Cont Batching + etc!

vMLX 1.6.29 delivers hybrid prefix caching for up to 20x faster follow-up turns and stable 100k-context multiturn sessions.

$ git clone --branch v1.6.29 https://github.com/jjang-ai/vmlx.git
# already have the repo? check out this version:
$ git checkout v1.6.29

• ›Hybrid prefix cache cuts follow-up turn latency on long documents by up to 20x (43.7k-token document follow-ups drop from 107s to 5s, byte-identical at temperature 0).
• ›100k-context multiturn sessions now proven stable: 97.6k-token conversations retrieve planted facts with 99.9% cache reuse and no memory faults.
• ›Chunked SSM re-derive ships enabled by default, restoring long-context cache reuse from 0% to 99.9% above 12.5k tokens on hybrid model families.
• ›Extends verified support across 12 model families, including DSV4 Flash native composite caching, Gemma 4 mixed-SWA with audio, Qwen 3.6/3.8 hybrid lines with native MTP, and TurboQuant KV bundles.