Pairing Claude Code with Native Fashions

# Introduction

 
Agentic coding classes are costly. A single Claude Code session — studying information, writing code, working checks, iterating — can burn 10–50x extra tokens than a plain chat dialog. At scale, that provides up quick. Add price limits that may interrupt a long-running workflow mid-session, and the dependency on a third-party API that may change pricing, implement stricter insurance policies, or go down at any level, and the case for native inference turns into easy.

Native fashions in 2026 are adequate. For the duties Claude Code handles every day — code completion, refactoring, debugging, codebase clarification — a well-chosen quantized mannequin working domestically covers the overwhelming majority of actual use circumstances at zero per-token price and with no price limits. This text covers three inference backends (Ollama, LM Studio, and llama.cpp), the precise setting variables and configuration information to wire every one to Claude Code, a curated desk of fashions price working, and the troubleshooting fixes for the problems you’ll really hit.

# How Claude Code Connects to Any Native Mannequin

 
The mechanism is easier than most guides make it look. Claude Code sends requests within the Anthropic Messages API format. By default these requests go to Anthropic’s servers. Setting ANTHROPIC_BASE_URL redirects them to any server that speaks the identical format, which now consists of Ollama, LM Studio, and llama.cpp natively.

In accordance with the official Claude Code setting variables documentation, the variables that matter for this setup are:

• ANTHROPIC_BASE_URL: redirects all API calls from Anthropic’s servers to no matter URL you set. Set this to your native inference server handle.
• ANTHROPIC_API_KEY: the API key despatched within the request header. Native servers usually ignore authentication, so that is normally set to a placeholder string like “native” or “ollama.”
• ANTHROPIC_AUTH_TOKEN: another auth header. Some native servers verify for this as an alternative of the API key. Set it to the identical placeholder.

ANTHROPIC_DEFAULT_SONNET_MODEL, ANTHROPIC_DEFAULT_HAIKU_MODEL, and ANTHROPIC_DEFAULT_OPUS_MODEL: Claude Code internally requests totally different mannequin tiers relying on the duty. These three variables map every tier to your native mannequin’s title. With out them, Claude Code sends requests for claude-sonnet-4-20250514 to your native server, which is able to reject the request as a result of no such mannequin exists domestically.

In January 2026, Ollama added native help for the Anthropic Messages API, which was the technical change that made this workflow sensible with out translation proxies. LM Studio added a local /v1/messages endpoint in model 0.4.1. llama.cpp has had direct Anthropic API help for longer. All three now converse Claude Code’s native protocol.

A clear structure diagram displaying Claude Code, Ollama, LM Studio, and llama.cpp | Picture by Writer

# Backend 1: Ollama

 
Ollama is the fitting place to begin. It handles all of the complexity of mannequin administration — downloading weights, quantization, GPU and CPU allocation, and serving — behind a easy command-line interface (CLI). One command to put in, one command to tug a mannequin, just a few setting variables to configure. It runs as a background service after set up, so there isn’t any handbook server begin required.

Conditions

• macOS, Linux, or Home windows (WSL2 beneficial on Home windows)
• No less than 16 GB RAM for sensible use (32 GB beneficial)
• GPU with 8+ GB VRAM for GPU inference, or CPU-only with sufficient RAM
• Ollama v0.14.0 or later required for Anthropic Messages API help

Set up Ollama:

# macOS and Linux -- one command set up
curl -fsSL https://ollama.com/set up.sh | sh

# Confirm the model -- have to be 0.14.0+ for Claude Code compatibility
ollama model
# Anticipated: ollama model is 0.14.x or increased

# Home windows: obtain the installer from https://ollama.com
# Native Home windows help has improved considerably in latest releases

After set up, Ollama begins routinely as a background service on port 11434. You possibly can confirm it’s working:

# Test the Ollama server is stay
curl http://localhost:11434

# Anticipated response:
# Ollama is working

Pull a coding mannequin:

# GLM-4.7-Flash -- beneficial place to begin
# Sturdy device calling, 128K context, matches on 8 GB VRAM
# Apache 2.0 license
ollama pull glm-4.7-flash:newest

# Qwen3-Coder -- sturdy code technology and instruction following
# Requires 20+ GB VRAM for the total mannequin
ollama pull qwen3-coder

# Devstral-Small -- particularly designed for agentic coding workflows
# Neighborhood-tested for Claude Code compatibility
# 24B, requires 16+ GB VRAM
ollama pull devstral-small-2:24b

# Confirm the mannequin is downloaded and prepared
ollama listing
# Reveals all pulled fashions with their sizes and modification dates

// Configuring Claude Code to Use Ollama

Choice 1: Shell export (present terminal session solely)

# Redirect Claude Code to your native Ollama server
export ANTHROPIC_BASE_URL="http://localhost:11434"

# Native servers don't require actual authentication
# Set these to any non-empty string -- Ollama ignores the worth
export ANTHROPIC_API_KEY="ollama"
export ANTHROPIC_AUTH_TOKEN="ollama"

# Map Claude Code's mannequin tier requests to your native mannequin title
# Claude Code internally requests sonnet/haiku/opus -- these variables
# translate these tier names to no matter mannequin you could have pulled domestically
export ANTHROPIC_DEFAULT_SONNET_MODEL="glm-4.7-flash:newest"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-4.7-flash:newest"
export ANTHROPIC_DEFAULT_OPUS_MODEL="glm-4.7-flash:newest"

# Launch Claude Code -- it is going to now use Ollama as an alternative of the Anthropic API
claude

Choice 2: ~/.claude/settings.json (everlasting, applies to all classes)

This strategy survives terminal restarts and applies each time you launch Claude Code. Claude Code reads setting variables from settings.json at startup in order that they take impact irrespective of how claude was launched.

Create or edit ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "http://localhost:11434",
    "ANTHROPIC_API_KEY": "ollama",
    "ANTHROPIC_AUTH_TOKEN": "ollama",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "glm-4.7-flash:newest",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "glm-4.7-flash:newest",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "glm-4.7-flash:newest"
  }
}

Choice 3: .env file in undertaking listing (per-project override)

If you’d like a selected undertaking to make use of a special mannequin whereas holding your world settings on the Anthropic API:

# .env in your undertaking root -- loaded routinely by Claude Code
ANTHROPIC_BASE_URL=http://localhost:11434
ANTHROPIC_API_KEY=ollama
ANTHROPIC_AUTH_TOKEN=ollama
ANTHROPIC_DEFAULT_SONNET_MODEL=qwen3-coder
ANTHROPIC_DEFAULT_HAIKU_MODEL=qwen3-coder
ANTHROPIC_DEFAULT_OPUS_MODEL=qwen3-coder

Confirm the connection:

# Launch Claude Code with a easy check
claude

# Inside Claude Code, run a fundamental immediate:
# > What mannequin are you working?
# A neighborhood mannequin ought to reply with out making any Anthropic API calls.

# To substantiate no exterior calls are being made, run with verbose logging:
claude --verbose

# Search for traces displaying requests going to localhost:11434
# quite than api.anthropic.com

Full working sequence from scratch:

curl -fsSL https://ollama.com/set up.sh | sh          # 1. Set up Ollama
ollama pull glm-4.7-flash:newest                       # 2. Pull mannequin (~4 GB)
export ANTHROPIC_BASE_URL="http://localhost:11434"     # 3. Redirect Claude Code
export ANTHROPIC_API_KEY="ollama"                      # 4. Set placeholder auth
export ANTHROPIC_AUTH_TOKEN="ollama"
export ANTHROPIC_DEFAULT_SONNET_MODEL="glm-4.7-flash:newest"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-4.7-flash:newest"
export ANTHROPIC_DEFAULT_OPUS_MODEL="glm-4.7-flash:newest"
claude                                                  # 5. Launch

# Backend 2: LM Studio

 
LM Studio is the fitting alternative if you would like a graphical interface for searching and managing fashions quite than working solely within the terminal. Since model 0.4.1, it features a native Anthropic-compatible /v1/messages endpoint — the identical path Claude Code expects — so no translation layer or proxy is required.

Conditions:

• macOS, Home windows, or Linux
• GPU with 6+ GB VRAM beneficial (CPU-only is feasible however sluggish)
• Obtain from lmstudio.ai or use the CLI installer for headless servers

Set up and configure LM Studio:

# On a server or VM with out a GUI -- CLI installer
curl -fsSL https://releases.lmstudio.ai/cli/set up.sh | bash

# Or obtain the desktop app from https://lmstudio.ai for GUI use

GUI setup steps:

1. Open LM Studio and seek for a coding mannequin (search “qwen coder” or “devstral”).
2. Obtain the mannequin. LM Studio handles quantization choice routinely.
3. Go to the Native Server tab (the <> icon within the left sidebar).
4. Set the context measurement. LM Studio recommends beginning with no less than 25,000 tokens and rising for higher outcomes.
5. Click on Begin Server.
6. Notice the port (default: 1234) and duplicate the mannequin title precisely as proven.

Notice: Copy the mannequin identifier precisely. LM Studio shows the precise string it is advisable cross to ANTHROPIC_DEFAULT_SONNET_MODEL. A mismatch right here is the commonest failure mode.

Configure Claude Code:

# Set the bottom URL to LM Studio's native server
export ANTHROPIC_BASE_URL="http://localhost:1234"
export ANTHROPIC_API_KEY="lm-studio"
export ANTHROPIC_AUTH_TOKEN="lm-studio"

# Substitute the mannequin title with what LM Studio exhibits to your loaded mannequin
# Copy it precisely -- together with any model suffix or quantization tag
export ANTHROPIC_DEFAULT_SONNET_MODEL="qwen2.5-coder-32b-instruct"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="qwen2.5-coder-32b-instruct"
export ANTHROPIC_DEFAULT_OPUS_MODEL="qwen2.5-coder-32b-instruct"

Or persistently in ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "http://localhost:1234",
    "ANTHROPIC_API_KEY": "lm-studio",
    "ANTHROPIC_AUTH_TOKEN": "lm-studio",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen2.5-coder-32b-instruct",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "qwen2.5-coder-32b-instruct",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen2.5-coder-32b-instruct"
  }
}

Find out how to run:

# 1. Begin the LM Studio server from the GUI (Native Server tab > Begin Server)
# 2. Set setting variables
export ANTHROPIC_BASE_URL="http://localhost:1234"
export ANTHROPIC_API_KEY="lm-studio"
export ANTHROPIC_AUTH_TOKEN="lm-studio"
export ANTHROPIC_DEFAULT_SONNET_MODEL="your-model-name-here"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="your-model-name-here"
export ANTHROPIC_DEFAULT_OPUS_MODEL="your-model-name-here"
# 3. Launch
claude

# Backend 3: llama.cpp

 
llama.cpp is the fitting alternative while you want direct management over inference parameters — quantization kind, KV cache configuration, batch measurement, thread rely — or if you find yourself working on a server and wish the bottom overhead. It has native Anthropic Messages API help, so no proxy or translation layer is required.

Conditions:

• A GGUF-format mannequin file (obtain from Hugging Face; seek for “GGUF” variations of any mannequin)
• CUDA-capable GPU for GPU inference, or CPU-only for slower inference
• CMake and a C++ compiler for supply builds (on Linux/CUDA, supply is beneficial)

Set up llama.cpp:

# macOS -- Homebrew is easiest
brew set up llama.cpp

# Linux with CUDA -- construct from supply for greatest GPU efficiency
git clone https://github.com/ggml-org/llama.cpp
cd llama.cpp
cmake -B construct -DGGML_CUDA=ON          # Allow CUDA acceleration
cmake --build construct --config Launch   # Construct
# Binaries in ./construct/bin/

# Linux CPU-only construct
cmake -B construct
cmake --build construct --config Launch

# Home windows -- pre-built binaries out there at:
# https://github.com/ggml-org/llama.cpp/releases
# Obtain the CUDA or CPU variant matching your {hardware}

Obtain a GGUF mannequin:

# Set up the Hugging Face CLI in the event you shouldn't have it
pip set up huggingface-hub

# Obtain GLM-4.7-Flash in Q4_K_XL quantization (~4.5 GB)
# This quantization presents an excellent measurement/high quality stability for coding
huggingface-cli obtain unsloth/GLM-4.7-Flash-GGUF 
  GLM-4.7-Flash-UD-Q4_K_XL.gguf 
  --local-dir ./fashions/

# Or obtain Qwen3-Coder in This autumn quantization (~15 GB for 32B)
huggingface-cli obtain Qwen/Qwen3-Coder-32B-Instruct-GGUF 
  qwen3-coder-32b-instruct-q4_k_m.gguf 
  --local-dir ./fashions/

Begin the llama.cpp server:

# Begin llama-server with Anthropic API help and a 128K context window
llama-server 
  --model ./fashions/GLM-4.7-Flash-UD-Q4_K_XL.gguf 
  --alias "glm-4.7-flash"           # This title goes in ANTHROPIC_DEFAULT_SONNET_MODEL
  --port 8001 
  --ctx-size 131072                 # 128K context -- necessary for big codebases
  --flash-attn                      # Reminiscence-efficient consideration, improves pace
  --n-gpu-layers 99                  # Offload all layers to GPU; take away for CPU-only

# For CPU-only inference (no GPU):
llama-server 
  --model ./fashions/GLM-4.7-Flash-UD-Q4_K_XL.gguf 
  --alias "glm-4.7-flash" 
  --port 8001 
  --ctx-size 32768                  # Cut back context measurement on CPU to maintain reminiscence manageable
  --threads 8                        # Match your CPU core rely

Key flags defined:

• --alias: the mannequin title string Claude Code will ship in requests. Set ANTHROPIC_DEFAULT_SONNET_MODEL to match this precisely.
• --ctx-size: context window in tokens. 131072 = 128K. Bigger is best for codebase evaluation however makes use of extra VRAM. Cut back in the event you get out-of-memory errors.
• --flash-attn: Flash Consideration reduces peak VRAM by processing consideration in smaller blocks. Allow it at any time when your construct helps it.
• --n-gpu-layers 99: offloads all transformer layers to the GPU. The server routinely makes use of fewer layers if VRAM is tight.

Configure Claude Code:

export ANTHROPIC_BASE_URL="http://localhost:8001"
export ANTHROPIC_API_KEY="llama-cpp"
export ANTHROPIC_AUTH_TOKEN="llama-cpp"

# Should match the --alias you handed to llama-server precisely
export ANTHROPIC_DEFAULT_SONNET_MODEL="glm-4.7-flash"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-4.7-flash"
export ANTHROPIC_DEFAULT_OPUS_MODEL="glm-4.7-flash"

Find out how to run:

# Terminal 1: begin the llama.cpp server
llama-server 
  --model ./fashions/GLM-4.7-Flash-UD-Q4_K_XL.gguf 
  --alias "glm-4.7-flash" 
  --port 8001 
  --ctx-size 131072 
  --flash-attn 
  --n-gpu-layers 99

# Terminal 2: configure and launch Claude Code
export ANTHROPIC_BASE_URL="http://localhost:8001"
export ANTHROPIC_API_KEY="llama-cpp"
export ANTHROPIC_AUTH_TOKEN="llama-cpp"
export ANTHROPIC_DEFAULT_SONNET_MODEL="glm-4.7-flash"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-4.7-flash"
export ANTHROPIC_DEFAULT_OPUS_MODEL="glm-4.7-flash"
claude

# The Full settings.json

 
Atmosphere variable exports final solely so long as the terminal session. For a sturdy configuration, use ~/.claude/settings.json. Claude Code reads variables from this file at startup in order that they apply irrespective of how Claude was launched — from the terminal, from a VS Code process, or from a script.

Here’s a production-ready settings.json with all variables defined:

{
  "env": {
    "ANTHROPIC_BASE_URL": "http://localhost:11434",

    "ANTHROPIC_API_KEY": "ollama",
    "ANTHROPIC_AUTH_TOKEN": "ollama",

    "ANTHROPIC_DEFAULT_SONNET_MODEL": "glm-4.7-flash:newest",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "glm-4.7-flash:newest",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "glm-4.7-flash:newest",

    "CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS": "1"
  }
}

Why CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS: "1" issues:

When utilizing Claude Code by way of non-Anthropic backends, Claude Code provides Anthropic-specific experimental beta flags to request headers — flags that third-party and native servers don’t acknowledge. This causes Error: Sudden worth(s) for the anthropic-beta header on most native inference servers. Setting this variable to "1" strips these headers earlier than the request goes out, which eliminates the error with out affecting any core Claude Code performance.

Switching between backends:

Should you work with a number of backends — Ollama for every day use, the Anthropic API for advanced duties — the cleanest strategy is sustaining separate shell scripts quite than modifying settings.json backwards and forwards:

# use-local.sh -- swap to Ollama
export ANTHROPIC_BASE_URL="http://localhost:11434"
export ANTHROPIC_API_KEY="ollama"
export ANTHROPIC_AUTH_TOKEN="ollama"
export ANTHROPIC_DEFAULT_SONNET_MODEL="glm-4.7-flash:newest"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-4.7-flash:newest"
export ANTHROPIC_DEFAULT_OPUS_MODEL="glm-4.7-flash:newest"
echo "Claude Code → native Ollama (glm-4.7-flash)"

# use-anthropic.sh -- swap again to the Anthropic API
unset ANTHROPIC_BASE_URL
unset ANTHROPIC_AUTH_TOKEN
unset ANTHROPIC_DEFAULT_SONNET_MODEL
unset ANTHROPIC_DEFAULT_HAIKU_MODEL
unset ANTHROPIC_DEFAULT_OPUS_MODEL
# ANTHROPIC_API_KEY ought to already be set to your actual key in your rc file
echo "Claude Code → Anthropic API"

Supply both script in your present session:

supply ./use-local.sh
claude

# Whenever you want the true API for a posh process:
supply ./use-anthropic.sh
claude

# Finest Native Fashions for Claude Code in 2026

 
{Hardware} is the principle constraint. For Claude Code with native fashions to be genuinely usable for coding duties quite than only a demo, purpose for 32 GB of RAM — Apple Silicon unified reminiscence or PC RAM. 16 GB is viable with smaller quantized fashions and CPU offload, however technology pace can be noticeably slower on multi-step agentic duties.

# Troubleshooting Frequent Points

1. Connection refused when launching Claude Code: The inference server will not be working. That is the commonest difficulty and the best to diagnose.

   # Test if Ollama is working
   curl http://localhost:11434
   # Anticipated: "Ollama is working"
   
   # Test if LM Studio server is working
   curl http://localhost:1234/v1/fashions
   # Ought to return a JSON listing of loaded fashions
   
   # Test if llama-server is working
   curl http://localhost:8001/well being
   # Ought to return {"standing":"okay"}
   
   # If not working -- begin the server first, then launch Claude Code
   ollama serve          # Ollama
   # LM Studio: use the GUI Native Server tab
   # llama.cpp: run the llama-server command from the Backend 3 part

    

2. Mannequin not discovered or unknown mannequin error: The mannequin title in your ANTHROPIC_DEFAULT_SONNET_MODEL doesn’t match what the server is aware of.

   # Checklist all fashions Ollama has out there
   ollama listing
   
   # The mannequin title in ANTHROPIC_DEFAULT_SONNET_MODEL should match EXACTLY
   # together with the tag -- "glm-4.7-flash:newest" not "glm-4.7-flash"
   
   # Confirm with a direct API name to substantiate what the server sees
   curl http://localhost:11434/v1/fashions

    

3. Instrument calls failing or returning errors: For streaming device calls, which Claude Code makes use of when executing features or scripts, Ollama model 0.14.3-rc1 or later is required. Earlier variations within the 0.14.x collection had incomplete streaming device name help.

   # Test your Ollama model
   ollama model
   
   # If beneath 0.14.3, replace Ollama
   curl -fsSL https://ollama.com/set up.sh | sh

    

4. anthropic-beta header error:

   You will note: Error: Sudden worth(s) for the anthropic-beta header. This occurs as a result of Claude Code provides Anthropic-specific experimental beta flags that native servers don’t acknowledge. Repair it by including this to your settings.json env block:

   "CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS": "1"

    

5. Reverting to the Anthropic API:

   # Shell session -- unset the redirect variables
   unset ANTHROPIC_BASE_URL
   unset ANTHROPIC_AUTH_TOKEN
   unset ANTHROPIC_DEFAULT_SONNET_MODEL
   unset ANTHROPIC_DEFAULT_HAIKU_MODEL
   unset ANTHROPIC_DEFAULT_OPUS_MODEL
   
   # Then be certain that your actual API key's set
   echo $ANTHROPIC_API_KEY
   # Ought to present your sk-ant-... key, not a placeholder
   
   # Should you used settings.json -- take away or remark out the env block
   # and restart Claude Code

    

6. Sluggish technology pace: For agentic Claude Code duties, technology pace issues as a result of every device name is a spherical journey. If pace is insufficient:
   • Change to a smaller or extra aggressively quantized mannequin (Q4_K_M as an alternative of Q8).
   • Allow --flash-attn in llama.cpp if not already set.
   • Cut back context measurement (--ctx-size); bigger contexts are slower to prefill.
   • On Ollama, set OLLAMA_NUM_GPU_LAYERS=99 in your setting to pressure most GPU offload.

# Conclusion

 
What used to require fragile adapters and hacks is now a five-step course of. Set up the inference backend, pull a mannequin, set three setting variables, and Claude Code routes to your native machine as an alternative of Anthropic’s API. The configuration takes below 5 minutes after you have the mannequin downloaded.

The sensible result’s a coding assistant that prices nothing to run after setup, has no price limits, retains your code solely in your machine, and covers the overwhelming majority of actual coding use circumstances at high quality ranges that weren’t out there in native fashions a yr in the past. Begin with Ollama and glm-4.7-flash — it has the bottom {hardware} requirement, probably the most constant tool-calling help, and the quickest path to a working setup. As soon as that’s working, scale up the mannequin primarily based in your {hardware} and the standard degree you really want.
 
 

Shittu Olumide is a software program engineer and technical author enthusiastic about leveraging cutting-edge applied sciences to craft compelling narratives, with a eager eye for element and a knack for simplifying advanced ideas. It’s also possible to discover Shittu on Twitter.