GPU memory allocation is unverified; external graphics hardware costs extra.
Listing evidence: 2026-09-24. Reatan manufacturer store, US plug and 48GB / 1TB selected, observed 25 Sep 2026 at USD 1299.99. The manufacturer marks this configuration sold out. Dated manufacturer reference only; Amazon price and availability are unverified. Tax and shipping may differ.
Fixed 36GB memory and internal storage; no CUDA support; price is a manufacturer reference rather than a verified Amazon offer.
Listing evidence: 2026-10-02. Apple US launch price for the base M5 Max Mac Studio, published 22 September 2026: USD 2,499. Dated manufacturer reference, not an Amazon offer; tax and shipping may differ.
Memory is soldered and shared with the operating system.
Listing evidence: 2026-09-24. GMKtec, EVO-X2 / 64GB / 1TB, recorded 22 Sep 2026; exact ASIN capacity identified 24 Sep. This is a dated reference, not a current Amazon price; tax, shipping and bundle contents may differ.
Weight size is not tested compatibility or speed. Check runtime support and memory for context before buying. Prices are dated references. As an Amazon Associate I earn from qualifying purchases.
How these options are selected
Up to three platforms, with the lowest dated reference price on each. Unknown prices follow known prices. Each memory pool keeps a reserve: 30% (minimum 2 GB) for system or unified memory; 20% (minimum 1 GB) for dedicated VRAM. System RAM is for CPU inference. This check does not establish GPU allocation, software support or context capacity.
Model file and details
Provider
Recursive
Total parameters
39,785,287,456
Exact file size
25,054,688,704 bytes
Quantization / precision
UD-Q4_K_XL
Weight repository
unsloth/flint-GGUF
Original model repository
RecursiveSI/flint
Weight revision
20468edfc38ab43d6914cc88b08cbecb05d0fc62
Model revision
2a247be0c0aec895ccb0ac7ca92229ec6df37288
Runtime notes
Confirm model architecture and quantization support in your runner.
Recorded release date
2026-10-21
Date evidence
Source repository created 21 October 2026; repository creation is an availability basis, not a proven announcement date. 39.8B-total, about 2.5B-active coding MoE; complete Unsloth GGUF file.
These scores were compiled from the linked sources. These published scores were not independently tested here. Different prompts, precision, thinking budgets and tools can change results.
SWE-bench Verified
74.2 / 100
Fixing real software issues with tools. Agent setup and external tools affect these scores. File size covers model weights only.
Benchmark efficiency: 2.962 published score points per GB of selected weights. This ratio is not an inference-speed measure; check the test settings to see whether the score covers the selected quantization.
Test settings: Recursive BF16 source evaluation with mini-swe-agent, temperature 1.0, top-p 0.95, top-k 20; average pass@1 over at least five attempts. Evaluator revision and uncertainty are not disclosed.
Publisher result, not an independent evaluation. Weight bytes exclude runtime and KV-cache memory. Maximum context does not establish device compatibility. Scores are for BF16, not this Unsloth quantization. Apache-2.0. Source SGLang kit targets B300/H200; GGUF architecture support must be checked in the selected runtime. No speed result is transferred from another device.
Agentic terminal-use task success. Publisher-reported percentages are retained, but agent scaffold, tools, timeouts and reasoning budget can materially affect results.
Benchmark efficiency: 2.191 published score points per GB of selected weights. This ratio is not an inference-speed measure; check the test settings to see whether the score covers the selected quantization.
Test settings: Recursive BF16 source evaluation: Terminus 2, temperature 1.0, top-p 0.95, top-k 20; average pass@1 over at least five attempts. Web-retrieved solutions count as unsuccessful.
Publisher result, not an independent evaluation. Weight bytes exclude runtime and KV-cache memory. Maximum context does not establish device compatibility. Scores are for BF16, not this Unsloth quantization. Apache-2.0. Source SGLang kit targets B300/H200; GGUF architecture support must be checked in the selected runtime. No speed result is transferred from another device.
Context and version evidence: The repository and file name identify the selected model. Revision hashes are shown above when recorded. Tested context length, full hardware configuration and run date are not consistently recorded in this snapshot. Only settings explicitly stated in each benchmark record are known.
The 25.055 GB figure covers model weights in this file. It excludes the context cache, runtime, operating system and any separate multimodal files. It is not a measured RAM requirement or a token-generation speed.
Confirm model architecture and quantization support in your runner. The selected weights use 25.055 GB; this excludes context cache, runtime buffers and other programs.
These are the first listed tiers that pass our reserve policy, not measured minimum requirements. System RAM and VRAM remain separate. Use the memory tool to change the reserve for your context and runner.