nowfound

AI · free · 2026

24 free alternatives to Qwen3.5

The 397B native multimodal agent with 17B active params

Below are 24 products that do a similar job and cost nothing to start: 17 free outright and 7 with a free plan and paid upgrades. Each price was read from the product's own site. They are ranked by how close each is in meaning, then by launch-day votes.

  1. 1

    Last week was big for open source LLMs. We got: - Qwen 2.5 VL (72b and 32b) - Gemma-3 (27b) - DeepSeek-v3-0324 And a couple weeks ago we got the new mistral-ocr model. We updated our OCR benchmark to include the new models. We evaluated 1,000 documents for JSON extraction accuracy. Major takeaways: - Qwen 2.5 VL (72b and 32b) are by far the most impressive. Both landed right around 75% accuracy (equivalent to GPT-4o’s performance). Qwen 72b was only 0.4% above 32b. Within the margin of error. - Both Qwen models passed mistral-ocr (72.2%), which is specifically trained for OCR. - Gemma-3…

    2025 · github.com · price checked November 2026

  2. 2
    Qwen3Free▲149

    Think Deeper or Act Faster

    2025 · github.com · price checked November 2026

  3. 3

    This project replicates mechanistic interpretability research on large language models using sparse autoencoders, building on published work from Anthropic, OpenAI, and DeepMind.

    2024 · github.com · price checked November 2026

  4. 4
    Qwen3.8-Flash-NextFree plan▲108

    The open-weight preview of Qwen4

    Aug 2026 · qwen.ai · price checked September 2026

  5. 5
    Qwen2.5-OmniFree▲165

    The end-to-end model powering multimodal chat

    2025 · github.com · price checked November 2026

  6. 6

    Metal-first MoE inference for Apple Silicon with bounded SSD expert streaming and local OpenAI Chat, Responses, and Anthropic Messages endpoints.

    Jul 2026 · github.com · price checked November 2026

  7. 7
    NeedleFree▲776

    Needle is a 26-million-parameter open-source model designed for function calling and tool use on consumer devices like phones and watches.

    May 2026 · github.com · price checked November 2026

  8. 8
    guppylmFree▲915

    Built a ~9M param LLM from scratch to understand how they actually work. Vanilla transformer, 60K synthetic conversations, ~130 lines of PyTorch. Trains in 5 min on a free Colab T4. The fish thinks the meaning of life is food. Fork it and swap the personality for your own character.

    Apr 2026 · github.com · price checked November 2026

  9. 9
    Z80-μLMFree▲514

    Z80-μLM is a character-level language model compressed into a 40KB file that runs on a Z80 processor with 64KB RAM, executable in CP/M emulators or on vintage hardware.

    Dec 2025 · github.com · price checked November 2026

  10. 10
    Conversational AI 2.0 From ElevenLabsFree plan, paid from $6/mo▲652

    Luke from ElevenLabs here. We just released Scribe Realtime v2, our new real-time Speech to Text model. We rebuilt it from the ground up around a new architecture that uses predictive transcription – it anticipates the next likely word before it’s spoken. That makes it both fast (around 150 ms latency) and highly accurate. Scribe v2 Realtime outperforms every low latency transcription model across 30 commonly used EU and Asian languages – 93.5% accuracy, compared to Gemini 2.5 Flash (91.4%), GPT-4o MiniTranscribe (90.7%), and Deepgram Nova 3 (85%). It supports 90+ languages including…

    2025 · elevenlabs.io · price checked November 2026

  11. 11
    DeepSeek-VL2Free▲503

    MoE vision-language, now easier to access

    2025 · github.com · price checked November 2026

  12. 12
    Qwen3-OmniFree▲137

    Native end-to-end multilingual omni-modal LLM

    2025 · github.com · price checked November 2026

  13. 13
    cogneeFree plan, paid from $99/mo▲382

    The complete guide to AI agent memory: long-term memory for LLM agents, memory architectures from vector stores to knowledge graphs, and how to build one.

    2025 · cognee.ai · price checked November 2026

  14. 14
    BeamFree▲136

    Highly efficient agentic 501B parameter open model

    Oct 2026 · reflection.ai · price checked November 2026

  15. 15
    Lamatic.aiFree plan, paid from $20/mo▲433

    Build AI agents in low-code & deploy on edge

    2024 · lamatic.ai · price checked November 2026

  16. 16

    Terminal-Bench-RL is an infrastructure for training reinforcement learning agents to perform complex terminal and coding tasks over long horizons.

    2025 · github.com · price checked November 2026

  17. 17
    VisionAgentFree plan, paid from $250/mo▲123

    Reasoning-Driven Agentic Object Detection

    2025 · landing.ai · price checked November 2026

  18. 18
    Activeloop AI Knowledge AgentFree plan, paid from $99/mo▲247

    Continual learning for every coding agent on your team.

    2025 · deeplake.ai · price checked November 2026

  19. 19

    Brain-inspired, multi-level reasoning & planning AI model

    2025 · github.com · price checked November 2026

  20. 20
    DeepSeek R1Free▲322

    Free, Fastest, 671B Full Version, No Sign-Up

    2025 · github.com · price checked November 2026

  21. 21
    Actian VectorAI DBFree plan, paid from $499/mo▲203

    The portable vector database for AI agents beyond the cloud

    Apr 2026 · actian.com · price checked November 2026

  22. 22

    Swiftlet is a Swift and Metal runtime that runs large Qwen Mixture-of-Experts models locally on Apple devices by streaming expert weights from storage, enabling 35B and 80B models to run with low RAM, including on iPhone.

    Aug 2026 · github.com · price checked November 2026

  23. 23
    LlamaGymFree▲239

    Fine-tune LLM agents with online reinforcement learning

    2024 · github.com · price checked November 2026

  24. 24

    An educational Rust project for exporting and running inference on Qwen3 LLM family

    2025 · github.com · price checked November 2026

Free tools by topic

All alternatives to Qwen3.5, paid too →About Qwen3.5 →

Free means the product’s own site says it is free or offers a free plan, as read on the date shown. Plans change; check the site before you rely on one.