Switchboard Router 4B has 4.022 billion total parameters. The selected Q8_0 GGUF file is 4.28 GB.
Data snapshot: . 1 published test in this catalog.
Model file and details
Provider
Switchboard
Total parameters
4,022,468,096
Exact file size
4,280,403,200 bytes
Quantization / precision
Q8_0
Weight repository
switchboardllm/switchboard-router-4b-GGUF
Original model repository
switchboardllm/switchboard-router-4b
Weight revision
6e0eca3db8b56f41fa739f26ff977c8eda417456
Model revision
1af75d6306014f57a7b24f49fb172a1190bf5207
Runtime notes
Confirm model architecture and quantization support in your runner.
Recorded release date
2026-10-13
Date evidence
The official Apache-2.0 source checkpoint and same-publisher Q8_0 GGUF were created on 13 October 2026. The source and GGUF cards document the same 10,000-prompt routing evaluation and a separate exact-Q8 fidelity check.
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.
Switchboard v0.1 · Tier accuracy
64 / 100
Publisher-reported exact three-tier routing accuracy on 10,000 held-out prompts from routellm/gpt4_dataset. This specialist routing result is not a general language-model quality score.
Benchmark efficiency: 14.952 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: Publisher evaluation of BF16 Switchboard 4B on the same 10,000 held-out routellm/gpt4_dataset prompts and three target tiers used for Switchboard 0.5B v0.1. Tier accuracy is the share matching the dataset-derived small, mid or frontier target tier.
Publisher result, not an independent evaluation and not a direct Q8 score. Labels derive from 2023 Mixtral-8x7B answers scored 1-5 by a GPT-4 judge rather than human preferences; the prompts are mostly English. The evaluation reports 95.0% served-well under its proxy, 5.0% sent too low and 0 invalid JSON outputs, but those are not catalog scores. The selected 4,280,403,200-byte Q8_0 GGUF was separately tested by the publisher with Ollama at temperature 0 on 347 held-out prompts: 339/347 tier decisions matched BF16, all 347 outputs were valid JSON and all domains matched. This fidelity sample does not establish exact-Q8 tier accuracy on the full 10,000 prompts. The model was trained on prompts cut to 512 tokens, reads one prompt rather than a conversation and must not be used as a security boundary because a prompt can influence its tier. The domain label is a keyword heuristic. Publisher timing and approximately 5 GB memory guidance are configuration-specific observations, not universal requirements. Apache-2.0 license.
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 4.28 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.
These complete configurations pass the initial weight-size check for the selected 4.28 GB file. Allow extra memory for context and software. Check the model’s runtime notes and your inference software before buying.
Listing evidence: 2026-09-25. Amazon.com base price visible in an owner-supplied screenshot reviewed 25 Sep 2026; capture time not supplied. Dated reference, not a live quote. Excludes card signup offers, coupons, tax and shipping.
Listing evidence: 2026-10-03. Apple US launch price announced 22 September 2026: USD 1,699. Dated manufacturer reference, not an Amazon offer; tax and shipping may differ.
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.
Memory matches are estimates, not tested compatibility or speed. As an Amazon Associate I earn from qualifying purchases.
Confirm model architecture and quantization support in your runner. The selected weights use 4.28 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.