# Wave 4 landscape — TML / Apple FM / code agents / Indic **Date**: 2026-05-31 **Question**: What's the competitive + capability landscape posttrainllm should know about before training specialists? **Outcome**: TML's "interaction model" framing maps onto Wave 2.6; Apple owns the same architecture but walled; code agents are cloud-dominated with a narrow local-first opening; Indic plan needs the `desi-max` reference replaced. ## 1. Thinking Machines Lab (TML) - **Founded Feb 2025** by Mira Murati + ex-OpenAI cohort (Schulman, Zoph, Weng, Tulloch, Metz). Raised $2B at $12B valuation in 5 months, reportedly chasing $50B by late 2025 ([Built In](https://builtin.com/articles/what-is-thinking-machines-lab), [Wikipedia](https://en.wikipedia.org/wiki/Thinking_Machines_Lab)). - **Tinker (Oct 2025)** — cloud distributed-training API for fine-tuning open-weight models (Qwen 4B-397B, Llama 1B-70B, DeepSeek V3.1, Kimi K2, Nemotron, GPT-OSS) via 4 primitives: `forward_backward`, `optim_step`, `sample`, `save_state`. LoRA-first; **not on-device** ([Tinker page](https://thinkingmachines.ai/tinker/), [DeepLearning.AI](https://www.deeplearning.ai/the-batch/thinking-machines-new-tinker-api-makes-it-easier-to-fine-tune-models-on-many-gpus)). - **Research blog Connectionism**: "Defeating Nondeterminism in LLM Inference" (Horace He), "LoRA Without Regret" (Schulman, 2025-09-29), "On-Policy Distillation" (2025-10-27), "Modular Manifolds" (Bernstein) ([blog index](https://thinkingmachines.ai/blog/)). - **"Interaction Models" thesis (May 2026)** — argues interactivity must be *native to the model*, with a **hybrid split**: a foreground "Interaction Model" doing 200ms micro-turns + a background model for async reasoning/tools ([interaction-models](https://thinkingmachines.ai/blog/interaction-models/)). - Posture: **cloud-only, infrastructure-focused, research-lab** — selling APIs to researchers, not shipping a consumer on-device product. **Implications for posttrainllm**: TML's interaction-model framing maps directly onto posttrainllm's Wave 2.6, but their implementation is cloud-bound. posttrainllm's differentiator is doing the foreground-interaction model *on-device* on Apple Silicon — that's a real differentiator, not duplication. Steal the architectural vocabulary ("foreground interaction model + background async") for the docs. Read "LoRA Without Regret" before more LoRA work. Do **not** try to compete with Tinker; if cloud fine-tune is ever needed, use it. ## 2. Apple Foundation Models + Private Cloud Compute - **~3B on-device model**: 5:3 block-depth split with KV-cache sharing (37.5% memory reduction) + 2-bit QAT; supports 65K context; **15 languages** via 150K-token vocab (up from 100K, only 25% more tokens) ([2025 updates](https://machinelearning.apple.com/research/apple-foundation-models-2025-updates), [tech report arXiv 2507.13575](https://arxiv.org/abs/2507.13575)). - **Server side**: Parallel-Track MoE (PT-MoE) cutting sync overhead 87.5%, running on Apple-silicon servers under Private Cloud Compute. - **Foundation Models framework (WWDC25)**: Swift API; `Tool` protocol handles guided generation + parallel/serial tool-call call graphs automatically; LoRA adapter fine-tuning supported via a Python toolkit ([WWDC25 #286](https://developer.apple.com/videos/play/wwdc2025/286/), [adapter training](https://developer.apple.com/apple-intelligence/foundation-models-adapter/)). - **PCC routing logic is NOT publicly documented** — the whitepaper covers attestation, secure enclave, hardened OS, but the "when to escalate" decision is opaque ([PCC blog](https://security.apple.com/blog/private-cloud-compute/)). - **Third-party model API access: none.** You get Apple's model via the Swift framework; you can ship LoRA adapters; you cannot substitute your own model into the framework or route your traffic through PCC. Adapters require re-training on every OS update. - Outperforms Qwen-2.5-3B across all 15 languages on internal evals; competitive with 4B models on English. **Implications for posttrainllm**: Apple owns the exact architecture posttrainllm is chasing, but the wall is real — closed model, closed router, adapter-only customization, OS-dependency on retraining. Position posttrainllm as the **open, hackable, multi-specialist** counterpart: - Can swap base models (Llama/Qwen/Sarvam) - Can do full SFT/DPO not just adapters - Can route to *any* cloud (not just PCC) - Targets devs/researchers Apple won't serve **Steal**: KV-cache sharing, the `Tool` protocol shape (already similar to the tool-call extractor plan), 2-bit QAT as a future quant target. **Don't** try to plug into App Intents — there's no public hook for third-party LLMs into Apple Intelligence. Stay parallel. ## 3. Code-agent architectures (Cursor / Continue / Cline / Aider) - **Cline** uses a **ReAct loop with structured-output enforcement** — rejects plain text, forces a tool call every turn (Plan-mode dialogue goes through a `plan_mode_respond` tool). Plan/Act split is the differentiator ([deepwiki](https://deepwiki.com/cline/cline/3.4-plan-and-act-modes), [GitHub](https://github.com/cline/cline)). SWE-bench Verified high-70s with Sonnet 4.5. - **Cursor Background Agent** + Cline both >59% on SWE-bench Verified with Claude Sonnet 4.6 — model dominates, the wrapper is a 3-5pt delta ([benchmark](https://awesomeagents.ai/leaderboards/swe-bench-coding-agent-leaderboard/)). - **Continue.dev** is the most local-friendly: first-class Ollama provider on `localhost:11434`, `provider: ollama` config, supports VS Code / JetBrains / Neovim ([docs](https://docs.continue.dev/customize/model-providers/top-level/ollama)). Lightweight middleware, no agent loop — closer to a "copilot" than an "agent." - **Aider** is terminal/git-first; "architect mode" splits planning model from editor model; uses **edit formats** (diff, diff-fenced, whole, editor-diff) as the structured output contract ([edit-formats](https://aider.chat/docs/more/edit-formats.html)). Architect mode 31.4% SWE-bench Verified — lower because of human-in- the-loop framing. - **None of them run a serious specialist model locally by default.** Local-LLM support exists but it's a degraded mode — they all assume cloud Sonnet/GPT for the real work. - Codeium/Windsurf excluded — fully cloud, closed. **Implications for posttrainllm**: The competitive gap is real but narrow. If posttrainllm ships a Mac dev tool *today* with on-device specialists, you lose on raw SWE-bench (Sonnet 4.6 > anything we fit in 3B). You win on: - **Latency**: sub-50ms TTFT vs 200-500ms cloud round-trip - **Cost**: zero per-token - **Privacy**: code never leaves device - **Multi-specialist routing**: none of them have it **Three concrete steals**: - Cline's structured-output-enforcement-via-tool (posttrainllm already does JSON mode — push harder) - Aider's edit-format contracts (cleaner than raw diff text for small models) - Continue's Ollama provider compatibility (ship a posttrainllm provider for Continue and you're instantly in dev workflows) Don't try to beat Cursor on SWE-bench; build the "local-first specialist with cloud escalation" framing they can't ship. ## 4. Multilingual / India focus - **desi-max is NOT a language model** — it's a 78-image LoRA on Qwen-Image-2512 for vintage South Asian *visual design* aesthetics ([HF card](https://huggingface.co/yenupam/desi-max)). Previously treated as a Hindi LLM base in `north_star_refined.md` — that was wrong; corrected 2026-05-31. - **Sarvam 30B and 105B** (open-sourced 2026, Apache 2.0): MoE with GQA (30B) / MLA (105B), 128 sparse experts each, 16T/12T pretrain tokens, custom tokenizer covering 22 scheduled Indian languages / 12 scripts. Government-selected for India's sovereign LLM via IndiaAI Mission ([Sarvam blog](https://www.sarvam.ai/blogs/sarvam-30b-105b), [sovereign LLM](https://www.sarvam.ai/blogs/indias-sovereign-llm)). Plan includes **Sarvam-Edge** for on-device — direct overlap with posttrainllm scope. - **AI4Bharat Airavata**: Hindi instruction-tuned LLM fine-tuned from OpenHathi using machine-translated English instruction sets via IndicTrans2; instruction datasets released publicly ([arXiv 2401.15006](https://arxiv.org/pdf/2401.15006), [IndicInstruct repo](https://github.com/AI4Bharat/IndicInstruct)). - **Krutrim** (Ola/Bhavish Aggarwal): covers 22 Indian languages but open-weight story is weaker than Sarvam ([Rest of World](https://restofworld.org/2026/india-frugal-ai-sarvam-krutrim-sovereign/)). - **MILU** (NAACL 2025, AI4Bharat): 8 domains, 41 subjects, 11 Indic languages, India-centric (regional exams, festivals, local history). 42 LLMs evaluated; GPT-4o leads at 74% ([MILU repo](https://github.com/AI4Bharat/MILU), [arXiv 2411.02538](https://arxiv.org/pdf/2411.02538)). - **IndicGenBench**: generative tasks across 29 Indic languages, extends Cross-Sum/XQuAD/XorQA/FLORES ([arXiv 2404.16816](https://arxiv.org/pdf/2404.16816)). - **Tokenizer reality check**: smollm2 and Qwen3 are NOT optimized for Devanagari. Sarvam's tokenizer is the right choice for serious Indic work; falling back to Qwen3 gives 2-4× token bloat on Hindi text. **Implications for posttrainllm**: Replace `desi-max` with **Sarvam-Edge (when released) or Airavata** as the Indic specialist starting point. The Indic specialist isn't a one-week dropin — tokenizer choice matters (likely retokenize with Sarvam's vocab or accept Qwen3 token-bloat penalty), and the eval harness needs MILU + IndicGenBench before claiming Hindi support. Sarvam is the most credible upstream — Apache-2.0 release + planned Edge variant means standing on their shoulders rather than training from scratch. Steal Airavata's translate-English-instructions-via-IndicTrans2 trick to bootstrap Hindi instruction data for specialists cheaply. ## Top 3 actions (next 3 months) 1. **Continue.dev provider adapter for posttrainllm** — Ollama-compatible endpoint that drops into Continue/Cline/Aider configs. Lowest-effort path to real users, validates the local-first thesis against the actual code-agent ecosystem, and provides a benchmark surface (SWE-bench mini) to track against Cursor/Cline. Pairs with the SSE streaming work already shipped. 2. **Adopt Apple's `Tool` protocol shape + Cline's structured-output- enforcement contract for the tool-call extractor** (Wave 2.6 mini- router). Don't reinvent — Apple's call-graph handling for parallel/serial tools is the right abstraction; Cline's "reject plain text, require tool call" is the right enforcement. Makes the eventual screen-reader specialist drop into a known-good shape. 3. **Fix the Indic plan**: replace desi-max with Sarvam-Edge / Airavata + wire MILU + IndicGenBench into the eval harness. Before any Hindi specialist training, run base Qwen3 / smollm2 through MILU to get a real baseline; that number tells you whether tokenizer-swap to Sarvam's vocab is worth the engineering cost. ## Key correction `desi-max` in `docs/roadmap/north_star_refined.md` was previously referenced as a Hindi LLM base. It is in fact a *text-to-image* LoRA on Qwen-Image-2512 for vintage Indian visual design. Replace with Sarvam-Edge (forthcoming) or Airavata. Corrected this revision. ## Sources - [TML Tinker](https://thinkingmachines.ai/tinker/), [TML Interaction Models](https://thinkingmachines.ai/blog/interaction-models/), [Connectionism blog](https://thinkingmachines.ai/blog/) - [Apple FM 2025 updates](https://machinelearning.apple.com/research/apple-foundation-models-2025-updates), [Apple FM tech report](https://arxiv.org/abs/2507.13575), [PCC](https://security.apple.com/blog/private-cloud-compute/), [WWDC25 #286](https://developer.apple.com/videos/play/wwdc2025/286/), [Apple adapter training](https://developer.apple.com/apple-intelligence/foundation-models-adapter/) - [Cline GitHub](https://github.com/cline/cline), [Aider edit formats](https://aider.chat/docs/more/edit-formats.html), [Continue+Ollama](https://docs.continue.dev/customize/model-providers/top-level/ollama), [SWE-bench leaderboard](https://awesomeagents.ai/leaderboards/swe-bench-coding-agent-leaderboard/) - [Sarvam 30B/105B](https://www.sarvam.ai/blogs/sarvam-30b-105b), [Airavata paper](https://arxiv.org/pdf/2401.15006), [MILU repo](https://github.com/AI4Bharat/MILU), [IndicGenBench paper](https://arxiv.org/pdf/2404.16816), [desi-max HF card](https://huggingface.co/yenupam/desi-max)