Instructions to use Voidstackloop/Tessera-125B-MoE with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Voidstackloop/Tessera-125B-MoE with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Voidstackloop/Tessera-125B-MoE:IQ2_S
# Run inference directly in the terminal:
llama cli -hf Voidstackloop/Tessera-125B-MoE:IQ2_S

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Voidstackloop/Tessera-125B-MoE:IQ2_S
# Run inference directly in the terminal:
llama cli -hf Voidstackloop/Tessera-125B-MoE:IQ2_S

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf Voidstackloop/Tessera-125B-MoE:IQ2_S
# Run inference directly in the terminal:
./llama-cli -hf Voidstackloop/Tessera-125B-MoE:IQ2_S

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf Voidstackloop/Tessera-125B-MoE:IQ2_S
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Voidstackloop/Tessera-125B-MoE:IQ2_S

Use Docker

docker model run hf.co/Voidstackloop/Tessera-125B-MoE:IQ2_S

Llama new
LM Studio
Jan
Ollama
How to use Voidstackloop/Tessera-125B-MoE with Ollama:
```
ollama run hf.co/Voidstackloop/Tessera-125B-MoE:IQ2_S
```
Unsloth Desktop

How to use Voidstackloop/Tessera-125B-MoE with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Voidstackloop/Tessera-125B-MoE:IQ2_S

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "Voidstackloop/Tessera-125B-MoE:IQ2_S"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Voidstackloop/Tessera-125B-MoE with Docker Model Runner:
```
docker model run hf.co/Voidstackloop/Tessera-125B-MoE:IQ2_S
```

Lemonade

How to use Voidstackloop/Tessera-125B-MoE with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Voidstackloop/Tessera-125B-MoE:IQ2_S

Run and chat with the model

lemonade run user.Tessera-125B-MoE-IQ2_S

List all available models

lemonade list

Hermes Agent

How to use Voidstackloop/Tessera-125B-MoE with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Voidstackloop/Tessera-125B-MoE:IQ2_S

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default Voidstackloop/Tessera-125B-MoE:IQ2_S

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Voidstackloop/Tessera-125B-MoE with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Voidstackloop/Tessera-125B-MoE:IQ2_S

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "Voidstackloop/Tessera-125B-MoE:IQ2_S" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Voidstackloop commited on 2 days ago

Commit

2d25512

verified ·

1 Parent(s): 557c8eb

Model card: benchmark results

Browse files

Files changed (1) hide show

README.md +34 -3

README.md CHANGED Viewed

@@ -16,8 +16,39 @@ Quantized, expert-pruned GGUF of Qwen/Qwen3.8-Flash-Next (architecture qwen4exp:
   `expert_count` = 448, `n_expert_used` = 10.
 * **Quantization** (`llama-quantize --pure` with a domain importance matrix): gate/up experts IQ2_S, down experts IQ4_NL, body Q6_K,
   per-layer n-gram table Q4_0 (loaded lazily with `-lzm on`), shared experts / hc / ple Q8_0.
-* `mtp-Qwen3.8-Flash-Next-balanced.gguf` is the multi-token-prediction draft head (not pruned), for `--spec-type draft-mtp`.
 * `SHA256SUMS` lists the checksums of both GGUF files.
-**Status at upload time: quality not measured yet.** Comparisons against the unsloth UD-IQ4_XS quant are being run; do not read anything into
-this file's quality until they are published. Needs a llama.cpp build with qwen4exp support (commit 7894e8b or newer).

   `expert_count` = 448, `n_expert_used` = 10.
 * **Quantization** (`llama-quantize --pure` with a domain importance matrix): gate/up experts IQ2_S, down experts IQ4_NL, body Q6_K,
   per-layer n-gram table Q4_0 (loaded lazily with `-lzm on`), shared experts / hc / ple Q8_0.
+* `Qwen3.8-Flash-Next-k448-IQ2_S.gguf` is 70.5 GiB; `mtp-Qwen3.8-Flash-Next-balanced.gguf` is the multi-token-prediction draft head (not pruned), for `--spec-type draft-mtp`.
 * `SHA256SUMS` lists the checksums of both GGUF files.
+* Needs a llama.cpp build with qwen4exp support (commit 7894e8b or newer).
+## Evaluation (2026-10-26)
+Everything below was run by the author on this exact file with llama.cpp. Each row names the setting; "best" means the best setting we tried for that
+benchmark, not a cherry-picked run, and every number comes from a single run. Samples are small (60 tasks per benchmark), so a 95% interval is
+roughly +/-12 points. Per-task tables are in `benchmarks/results/`, the scripts in `benchmarks/scripts/`, the full write-up in `benchmarks/REPORT.md`.
+| Benchmark | Best result | Setting | Notes |
+|---|---|---|---|
+| **LiveCodeBench v6** (60 of 175 problems, 2025-01 to 2025-04), best | **54/60 = 90.0%** | thinking, 16,384-token budget + up to 2 repair turns | **Chained estimate, not one clean run**: 41 passes from the 4k run + 9 rescued by rerunning the 21 failures at 16k + 4 rescued by repair turns. Repair turns show the model the first failing visible sample test (hidden tests only score the result), so this is an agent-style loop, not single-shot. |
+| LiveCodeBench v6, clean single attempt | 49/60 = 81.7% | thinking, 16,384-token budget, one attempt, all 60 rerun | The comparable number for single-shot leaderboards. 4k budget: 41/60 (68.3%); no-think: 38/60 (63.3%). 16k vs 4k paired p = 0.021. |
+| CRUXEval (60 tasks) | **58/60 = 96.7%** | any setting | At its ceiling in every run. |
+| Coding total (LiveCodeBench + CRUXEval, 120 tasks) | **112/120 = 93.3%** (estimate, with repair turns); clean single attempt 107/120 = 89.2% | thinking, 16k budget (+ repair turns for the 112) | 4k: 99/120; no-think: 96/120. |
+| HumanEval+ / MBPP (60 tasks) | 55/60 = 91.7% | no-think, greedy | HumanEval+ 22/24, MBPP 33/36. |
+| MATH-500 (45 tasks) | 39/45 = 86.7% | no-think, greedy, 2,048-token cap | AIME 0/15 in the same run: every answer hit the token cap, so it is not a measure of ability. Thinking mode on math was only tested on a few failures. |
+| MMLU-Pro (40) + BBH (20) | 50/60 = 83.3% | no-think | BBH 20/20, MMLU-Pro 30/40. Thinking with a 4k budget rescued 5 of the 10 failures (about 55/60 as an estimate); a 16k budget rescued 3 of 10. |
+| English (RACE, CoLA, rule-following; 60 tasks) | 52/60 = 86.7% | no-think | Quick set; not compared with other models. |
+| Agentic: SWE-bench Verified Mini (20 runnable instances) through Claude Code | **7/20 = 35%** with no web lookups; 15/20 as run | no-think, 65k context | **The 15/20 is not valid**: the sandbox had internet and the agent fetched upstream fixes in 8 of the 15 resolved instances. 7/20 counts only the instances solved without any lookup (a floor). An offline rerun has not been completed. |
+### Against the unsloth UD-IQ4_XS quant (87 GiB)
+No-think greedy on a 120-task suite (HumanEval+, MATH-500, MMLU-Pro; 40 each): Q4 103/120, this model 101/120, paired sign test p = 0.754.
+**It matches Q4 within noise; it does not beat it.** Same suite on the same machine: 11.7 vs 8.2 tokens/s, and the file is 17 GiB smaller.
+The full-precision parent was not run by us; vendor-reported LiveCodeBench v6 for the parent is 91.9.
+### Speed on an RTX 4060 Laptop (8 GB) with 96 GB RAM, native Windows llama.cpp with a GPU expert cache
+About 22 tokens/s decoding short code answers (MTP draft head, or a large GPU expert cache without MTP); 18 to 19 tokens/s with a 128k context
+window configured; 14 to 16 tokens/s for prose. Prefill about 125 to 165 tokens/s. Measured with no game running; a running game cut it to 12 tokens/s.
+### Caveats
+* One run per setting, 60 tasks per benchmark, a few problems of run-to-run noise.
+* The thinking-budget results are fixed-budget runs; the hard LiveCodeBench problems use the whole budget.
+* Competitor numbers are not included because we did not run other models ourselves.
+* This is an experimental, pruned, 2-bit-class quantization: expect weaker results than the full-precision parent on hard tasks.