--- title: "Claude Code Backends Compared — ToolPiper vs Ollama vs vLLM vs LM Studio" description: "Side-by-side comparison of every Anthropic-compatible local backend for Claude Code. Features, performance, BYOK support, and where each one wins." canonical: "https://modelpiper.com/compare/claude-code-backends/" --- # Claude Code Backends Compared — ToolPiper vs Ollama vs vLLM vs LM Studio > Side-by-side comparison of every Anthropic-compatible local backend for Claude Code. Features, performance, BYOK support, and where each one wins. [![ModelPiper](model-piper-logo-sm.svg)ModelPiper](/) [ToolPiper](/toolpiper)[Docs](/docs/toolpiper)[Download ToolPiper](https://modelpiper.com/download) Comparison · Updated 2026-04-28 # The best local backend for Claude Code. Claude Code runs against any HTTP endpoint that speaks the Anthropic Messages API. Four serious options: ToolPiper, Ollama, vLLM, LM Studio. They are not equivalent. Here is what is actually different. ## The short version - **Pick Ollama / vLLM / LM Studio** if you want a single local model server and nothing else. They speak the protocol; that is the feature. - **Pick ToolPiper** if you want Claude Code to route across local + cloud providers in the same conversation, with 147 MCP tools, voice, browser automation, and Keychain-locked BYOK — without configuring any of it. - ToolPiper is the only backend that ships _conversational provider switching_ ("use my local one") and _context-aware recommendations_ ("you're at 47K, switch to Gemini Flash"). That is the moat, not raw inference speed. ## Feature matrix Feature ToolPiper Ollama vLLM LM Studio Anthropic Messages API (\`POST /v1/messages\`) Native, with provider switching No Yes (since v0.7) Yes (since 0.3.x) OpenAI Chat Completions (\`/v1/chat/completions\`) Yes Yes Yes Yes Streaming with cancellation <200 ms upstream abort on Esc Yes Yes Yes Tool use round-trip Anthropic + OpenAI shapes OpenAI only OpenAI only OpenAI only Inference backends 9 (llama.cpp, Apple Intelligence, FluidAudio, MLX, Apple Vision, …) 1 (llama.cpp) 1 (vLLM engine) 1 (llama.cpp) Apple Intelligence (Neural Engine) backend Yes — first to ship it No No No BYOK cloud proxy (OpenAI / Anthropic / Gemini / etc.) Yes — Keychain-locked, server-side injection No No No Cross-provider routing in one base URL Yes — \`endpoint\_set\` MCP tool No No No Conversational provider switching "Switch to my local one" works in chat No No No Context-aware backend recommendation \`endpoint\_recommend\` reads conversation size + telemetry No No No Native /model picker integration Yes — auto-writes \`ANTHROPIC\_CUSTOM\_MODEL\_OPTION\_\*\` No No No Zero-config Claude Code install One click — token + settings.json + MCP register Manual env vars Manual env vars + DEFAULT\_\*\_MODEL dance Manual env vars + DEFAULT\_\*\_MODEL dance MCP server bundled 147 tools — browser, files, system, video, voice No No No MCP toolbelt injection (when client has no tools) Header-gated, PiperMatch-retrieved No No No Browser automation (CDP) 17 AX-native tools No No No Voice AI (STT + TTS + voice clone) Yes No No No OCR (Apple Vision) Yes No No No Image / video upscale on Neural Engine PiperSR — 44 fps real-time 2× video No No No Resource intelligence (auto model eviction) Real-time RAM monitoring + eviction Reads RAM once at startup Manual Removed in 4.0 Distribution Native macOS DMG Homebrew + CLI Python wheel + GPU container Cross-platform DMG Sandboxing Native, signed No No No Price Free; Pro $9.99/mo for cloud BYOK Free Free Free ## Performance baselines Performance numbers depend heavily on your Mac, your model, and your prompt mix. ToolPiper ships `scripts/anthropic-perf-baseline.ts` — a harness that captures p50 / p95 TTFT and tokens/sec across _every_ provider you have configured, including Ollama / vLLM / LM Studio if you wire them as endpoints. The output lives at [docs/features/anthropic-perf-baseline.md](https://github.com/ModelPiper/modelpiper/blob/main/docs/features/anthropic-perf-baseline.md) and updates on every run. The numbers below are workload definitions; the live comparison is in the repo. Workload ToolPiper Definition Single-turn 2K prompt (TTFT, p50) See live numbers Captured per \`EndpointConfig\` by \`scripts/anthropic-perf-baseline.ts\`. Runs across every provider you have configured. Multi-turn 20K with tools (tokens/sec, p50) See live numbers Same harness — Phase 7 deliverable, populated on every fresh ToolPiper run. Tool-heavy 50K (TTFT, p95) See live numbers Skipped automatically on backends whose context window is below 50K — annotated \`context\_overflow\` in the output. ## Pick a recipe [Free · On-device ### Apple Intelligence Neural Engine inference. No download, no quota, no API key. The cheapest fastest path. ](/recipes/claude-code-apple-intelligence)[Local · 32K · Tool use ### Local Qwen Qwen-Coder on llama.cpp via Metal GPU. Best capability/privacy balance. ](/recipes/claude-code-local-qwen)[BYOK · Keychain ### OpenAI with your key gpt-4o through Claude Code, billed against your account, key locked in Keychain. ](/recipes/claude-code-byok-openai)[Conversational ### Switch providers in chat "Use my local one." "Try Gemini Flash for this." No env vars. The MCP-native flow. ](/recipes/claude-code-cross-provider) ## Try it in two minutes Download ToolPiper, click "Configure for Claude Code", run `claude`. Done. [Download ToolPiper](https://modelpiper.com/download)[Open Setup Guide](/docs/toolpiper#claude-code)