Model reference · open weights

Qwen3.5-MTPLX-Optimized-Speed

Available as managed deployment LLMs Youssofal · community Text gen 1 variants 2k dl/mo

Qwen3.5-MTPLX-Optimized-Speed is an open-weight language model from Youssofal. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.

Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.

What it is

Released byYoussofal
TypeLanguage models
TaskText gen
Parameters (lead)9.4B
Context256k tokens
Runs withmlx
Released2026-06-07
Popularity2k downloads / month
LicenceOpen weights

About

What Qwen3.5-MTPLX-Optimized-Speed is

Fast local 9B inference for Apple Silicon, packaged for MTPLX native Multi-Token-Prediction speculative decoding.

This is the 9B speed checkpoint: a 6-bit MLX body with BF16 MTP heads, tuned as the stronger small-Mac option above the 4B release.

Read the full model card

Run It

brew install youssofal/mtplx/mtplx
mtplx start
mtplx run "hello" --model Youssofal/Qwen3.5-9B-MTPLX-Optimized-Speed

For an OpenAI-compatible local server:

mtplx serve --model Youssofal/Qwen3.5-9B-MTPLX-Optimized-Speed --profile sustained --max --port 8000 --no-stats-footer

Why This Exists

MTPLX uses the model's own MTP heads to generate draft tokens, then verifies those tokens with the main model. When the draft heads are well-matched, you get higher throughput without running a separate drafter model.

MTPLX reads mtplx_runtime.json and selects the measured defaults automatically.

Recommended Runtime Defaults

SettingValue
Backendqwen3-next-mtp
Default depthD2
Target samplertemp=0.60, top_p=0.95, top_k=20
Draft samplertemp=0.60, top_p=0.95, top_k=20
Profilesustained
Benchmark fan modemax

Performance

Measured in MTPLX on Apple Silicon using the release runtime path.

ModeTPSVerify timeAcceptance
AR baseline64.96--
D1 comparison92.876.83s0.9120
D2 promoted default101.324.13s0.9398, 0.8102
D3 comparison96.304.58s0.9278, 0.7732, 0.6443

Model Build

ComponentFormat
Main body6-bit MLX affine, group size 64
MTP headsBF16 native MTP sidecar
Samplertarget and draft temp=0.60, top_p=0.95, top_k=20

This is not a full-precision checkpoint. It is built for fast local use on Apple Silicon through MTPLX.

Files

  • model-*.safetensors: MLX body shards
  • mtp.safetensors: MTP sidecar
  • mtplx_runtime.json: MTPLX runtime contract and measured defaults
  • MTPLX_PUBLISH_MANIFEST.json: file sizes and artifact metadata
  • tokenizer and config files for local loading

From the published model card. Full card on the HuggingFace links in the sidebar.

Using it via the API

Call it like any OpenAI endpoint

Once AxForge deploys qwen3-5-mtplx-optimized-speed for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (qwen3-5-mtplx-optimized-speed below is illustrative; you get the exact model name on deployment.)

$ curl -sS https://api.axforge.ai/v1/chat/completions \
  -H "Authorization: Bearer $AXFORGE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"qwen3-5-mtplx-optimized-speed","messages":[{"role":"user","content":"Hello"}]}'

Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.

© 2026 AxForge · EU-hosted AI infrastructure Pricing Docs Trust Privacy Terms