mudler/LocalAI · critical

failed loading model (generic error)\n

Error message

failed loading model (generic error)\n

What it means

Fatal load error: new_sd_ctx() returned NULL, meaning the upstream stable-diffusion.cpp library refused to create a context. The shim prints this generic message (upstream's own, more specific errors appear earlier on stderr) and returns 1, aborting the load. Typical root causes are unreadable/corrupt model files, missing VAE/CLIP components, unsupported architecture, or out-of-memory during weight loading.

Source

Thrown at backend/go/stablediffusion-ggml/cpp/gosd.cpp:676

    ctx_params.diffusion_flash_attn = diffusion_flash_attn;
    ctx_params.tae_preview_only = tae_preview_only;
    ctx_params.diffusion_conv_direct = diffusion_conv_direct;
    ctx_params.vae_conv_direct = vae_conv_direct;
    ctx_params.force_sdxl_vae_conv_scale = force_sdxl_vae_conv_scale;
    // Chroma knobs: upstream dropped the dedicated chroma_use_dit_mask /
    // chroma_use_t5_mask / chroma_t5_mask_pad struct fields and now reads them
    // from the generic model_args key=value spec (parse_key_value_args). Emit
    // them there so the existing chroma options keep working. This string must
    // outlive new_sd_ctx() below.
    std::string model_args_spec =
        "chroma_use_dit_mask=" + std::string(chroma_use_dit_mask ? "true" : "false") +
        ",chroma_use_t5_mask=" + std::string(chroma_use_t5_mask ? "true" : "false") +
        ",chroma_t5_mask_pad=" + std::to_string(chroma_t5_mask_pad);
    ctx_params.model_args = model_args_spec.c_str();
    sd_ctx_t* sd_ctx = new_sd_ctx(&ctx_params);

    if (sd_ctx == NULL) {
        fprintf (stderr, "failed loading model (generic error)\n");
        // TODO: Clean up allocated memory
        return 1;
    }
    fprintf (stderr, "Created context: OK\n");

    int sample_method_found = -1;
    sample_method_t sm = str_to_sample_method(sampler);
    if (sm != SAMPLE_METHOD_COUNT) {
        sample_method_found = (int)sm;
        fprintf(stderr, "Found sampler: %s\n", sampler);
    }
    if (sample_method_found == -1) {
        sample_method_found = sd_get_default_sample_method(sd_ctx);
        fprintf(stderr, "Invalid sample method, using default: %s\n", sd_sample_method_name((sample_method_t)sample_method_found));
    }
    sample_method = (sample_method_t)sample_method_found;

    scheduler_t sched = str_to_scheduler(scheduler_str);

View on GitHub (pinned to 44413a9d06)

Solutions

1. Read the upstream stderr lines immediately above this message — they name the actual failing file or reason
2. Verify every configured path (model, vae, taesd, control_net, etc.) exists and is a supported format
3. Check free RAM/VRAM and try a quantized wtype or smaller model
4. Re-download the model if the file is truncated/corrupt (compare size/checksum with the source)

Example fix

# before: model file missing
options: "model=/models/sdxl.safetensors,vae=/models/vae.safetensors"
# -> failed loading model (generic error)

# after: verify paths exist
ls -la /models/sdxl.safetensors /models/vae.safetensors
options: "model=/models/sdxl.safetensors,vae=/models/vae.safetensors"

// Go caller: verify every configured model file before calling load
paths := []string{opts.Model, opts.Vae, opts.Taesd, opts.ControlNet}
for _, p := range paths {
    if p == "" { continue }
    if fi, err := os.Stat(p); err != nil || fi.Size() == 0 {
        return fmt.Errorf("model file %q missing or empty", p)
    }
}
// rough memory pre-check: weights ~ file size * (1..2)
if m := opts.ExpectedWeightMemoryGB; m > 0 && memAvailableGB() < m {
    return fmt.Errorf("insufficient memory: need ~%dGB", m)
}

When it happens

Trigger: Calling load with model_path/vae_path/clip paths that do not exist or are corrupt; passing a Flux/Chroma pipeline missing its diffusion-model or text-encoder components; requesting a quantization that fails on this hardware; exhausting RAM/VRAM while loading weights.

Common situations: Wrong path in gallery YAML; partially downloaded .safetensors/.gguf; using a diffusers-only model without converting; small machines loading large fp16 models; missing embeddings_connectors/pulid/control-net files referenced in options.

Related errors

• [acestep-cpp] FATAL: failed to load condition encoder\n
• Failed to allocate memory for resized image\n
• model snapshot does not exist: {model_ref}
• model snapshot must contain exactly one {suffix} file; found
• model_id is required to load a pipeline

AI-assisted analysis of mudler/LocalAI@44413a9d06 (2026-08-15). Data as JSON: /api/errors/704521cd57074cd2. Report an issue: GitHub.