Downloads · 30 days
145
29% of all-time downloads
needmorevramog/Tinfield-1-Q4_K_XL-GGUF
Tinfield-1-Q4_K_XL-GGUF is a image-text-to-text model from needmorevramog. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as other.
This repository contains a Q4KXL-style GGUF quantization of Tinfield 1, an agentic model for terminal work and long-horizon software engineering.
Downloads · 30 days
145
29% of all-time downloads
All-time downloads
499
Public
Repo size
111 GB
Likes
1
Public
Click a slice to open those files.
.gguf111 GB · 100%
From the Hugging Face model README
This repository contains a Q4_K_XL-style GGUF quantization of Tinfield 1, an agentic model for terminal work and long-horizon software engineering.
The quant was produced from the Tinfield 1 Q8_0 GGUF using llama.cpp and the official Unsloth Qwen3.8-Flash-Next importance matrix.
| Value | |
|---|---|
| Source | Tinfield 1 Q8_0 |
| Output size | ~111.3 GB |
| Binary size | ~103.6 GiB |
| Effective precision | 5.03 BPW |
| Quantization | Mixed Q4_K / Q5_K / Q5_1 / Q8_0 / IQ4_NL |
| Importance matrix | Unsloth Qwen3.8-Flash-Next |
| Format | GGUF |
| Runtime | llama.cpp compatible |
This is a community quantization and is not an official Unsloth release.
Tinfield 1 scores 33.0 on Terminal-Bench 4.0, ahead of Claude Opus 4.8 at 23.6 and Claude Sonnet 5 at 12.4, and 62 on DeepSWE v1.1, ahead of Qwen3.8 Max, DeepSeek V4 Flash and Opus 4.8 again. On the Terminal-Bench board it is the second open-weight model, on a leaderboard otherwise made up entirely of frontier closed models.
It does that on 6.6B active parameters per token out of 177B total.
Built on Qwen3.8-Flash-Next. 256k context.
This build was requantized from a Tinfield 1 Q8_0 GGUF rather than directly from the original BF16 checkpoint.
The quantization uses the importance matrix published by Unsloth for its Qwen3.8-Flash-Next UD quants:
unsloth/Qwen3.8-Flash-Next-GGUF/imatrix_unsloth.gguf_file
| Tensor group | Quantization |
|---|---|
blk.2.ffn_gate_exps / blk.2.ffn_up_exps | Q5_K |
Other ffn_gate_exps / ffn_up_exps | Q4_K |
ffn_down_exps in blocks 2, 4, 30, 46, 47 | Q8_0 |
Other ffn_down_exps | Q5_1 |
per_layer_token_embd | IQ4_NL |
| Remaining quantizable tensors | Q8_0 |
This corresponds to approximately:


The following benchmark scores are from the original Tinfield 1 model and are not measurements of this Q4_K_XL quant.
| Benchmark | Tinfield 1 | Qwen3.8-Flash-Next |
|---|---|---|
| Terminal-Bench 4.0 | 33.0 | 29.0 |
| DeepSWE v1.1 | 62.0 | 58.7 |
Evaluated with mini-swe-agent at k=5 on the full task sets: 66 tasks for Terminal-Bench 4.0, 113 for DeepSWE v1.1.
| Model | Agent | Score |
|---|---|---|
| GLM-5.3 (max) | Claude Code | 41.8 |
| GPT-5.6 Sol (max) | Codex | 37.3 |
| Tinfield 1 | mini-swe-agent | 33.0 |
| Qwen3.8-Flash-Next | 29.0 | |
| Claude Opus 4.8 (max) | Claude Code | 23.6 |
| GPT-5.6 Terra (max) | Codex | 21.5 |
| Grok 4.6 (high) | Grok Build | 20.3 |
| Gemini 3.8 Flash (high) | mini-swe-agent | 19.1 |
| Claude Sonnet 5 (max) | Claude Code | 12.4 |
| Model | Score |
|---|---|
| GLM-5.3 (max) | 69 |
| GLM-5.3 Flash (max) | 63 |
| DeepSeek V4 Pro (max) | 63 |
| Tinfield 1 | 62 |
| Claude Opus 4.8 (max) | 59 |
| Qwen3.8 Max (xhigh) | 57 |
| Muse Spark 1.2 (xhigh) | 55 |
| Claude Sonnet 5 (max) | 54 |
| DeepSeek V4 Flash (max) | 53 |
Comparison figures from the Terminal-Bench and DeepSWE leaderboards. Base model Terminal-Bench figure from Artificial Analysis; base model DeepSWE figure from the Qwen3.8-Flash-Next model card.
Qwen Community License 1.0, following the upstream Tinfield 1 model.