Cohere Transcribe 03-2026 vs Granite Speech 5.0 2B

Cohere Transcribe 03-2026 65

Memory earns 30, since the BF16 file is 4.13 GB and needs about 5 GB. Runtimes get 10 for Transformers (v5.4.0 to v5.22.1) and vLLM (v0.32.0 and v0.33.0), both released; llama.cpp, SGLang, Ollama and LM Studio don't run it. Builds score 0, as Cohere publishes no quantised build. Apache 2.0 with nothing added gives terms 15. Download gets 5 because the repository is a click-through gate that needs an account. Hosted too gets 5 for Cohere's own API and fal.

Granite Speech 5.0 2B 95

Memory 30, the smallest build listed that keeps 4 bits is IBM's Q4_K_M GGUF, 1.14 GB with the 1.16 GB audio projector, and 2.30 GB × 1.2 = 2.8 GB, within 16 GB. IBM recommends Q8_0, 3.12 GB with the projector. Runtimes 20, Released Transformers 5.22.2 and vLLM 0.33.0 have the granite_speech model, llama.cpp b12242 has its converter and audio front end, and Ollama carries it as ibm/granite5-speech, which the card says was tested with v0.40.15. Builds 15, IBM publishes GGUF builds at fourteen quantised precisions, from Q2_K to Q8_0, with the audio projector. Terms 15, Apache 2.0 in the metadata and on the card, with nothing added. Download 10, the Hub API reports gated false for both repositories on 10 November 2026, so the files download with no account and nothing to accept. Hosted 5, IBM runs a Hugging Face Inference Endpoint that serves this model through an OpenAI-compatible API, with the address and code on the card. 30 + 20 + 15 + 15 + 10 + 5 = 95.

Cohere Transcribe 03-2026 64

The public card covers intended use, limitations, architecture, parameters and languages, without a context length or cut-off (14). Training data has a size and method but no sources or cut-off (5). The last change is the card commit of 7 October 2026 (pull request 49, which adds runtime and app support notes), 28 days before 4 November; card and documentation commits count on this line in this re-check, as for other open-weight listings (20; 10 before, on the 28 April configuration file). Cohere answered 11 of 23 questions, none since April (10). It's a first release with a dated name (5), and every report of a broken file has a reply (10). 14 + 5 + 20 + 10 + 5 + 10 = 64.

Granite Speech 5.0 2B 92

Model card 17, Intended use, limitations, the architecture, the size, the 16K positions and two-minute segments, and the languages are on the card. Training data 20, the data is listed with sources and sizes: a 39-row table giving each corpus, its task, its hours and a link to where it comes from, with the synthetic sets described. Last change 20, the last commit, on 6 November 2026, updated the card, 4 days before this check. Responsiveness 20, Both discussions that aren't pull requests (21 and 28 October 2026) have a reply from an ibm-granite member. Version policy 5, granite-speech-4.1-2b stays in its own repository and 5.0 came out in a new one, which names its predecessor and gives a release date. File health 10, Config, template and card agree on the architecture and the 16K positions. 17 + 20 + 20 + 20 + 5 + 10 = 92.

Cohere Transcribe 03-2026 50

File safety gets 15, safetensors only but served in vLLM with --trust-remote-code as the card documents. Provenance gets 10 for Cohere's verified organisation; the checksums are hidden behind the gate and weren't seen. Licence risk keeps 15 under Apache 2.0. Agent guidance gets 5 for cautions about hallucinating on noise. The security programme gets 5 for Cohere's disclosure policy and bug bounty, with no security.txt and no public advisories. Safety evaluation scores 0.

Granite Speech 5.0 2B 70

File safety 25, safetensors and GGUF only, no .bin, .pt or .pth files and no Python in either repository. Provenance 15, the files sit in the ibm-granite organisation, which IBM's Granite docs link to (10). Safety evaluation 0, no safety results on the card. Agent guidance 10, the card says what the model does with a prompt it doesn't know: it ignores an unfamiliar or malformed prompt and transcribes, which it says limits how audio can steer the system. Licence risk 15, Apache 2.0 with no use policy, no threshold and no right to change the terms. Security programme 5, www.ibm.com/.well-known/security.txt names IBM PSIRT, read for the Granite 4.2 3B listing on 4 November 2026 (5). 25 + 15 + 0 + 10 + 15 + 5 = 70.

Cohere Transcribe 03-2026 45

Read for a speech model, the tool-calling, trained tool calling and structured-output lines stand for what it takes and returns. Cohere documents the audio, language and punctuation inputs and the decoded text with chunk reassembly in Transformers, and the vLLM transcription request (25). It was trained for one stated language per clip (10). There are no timestamps, segments or documented JSON (0). The chat template line gets 10 because the prompt is built by the processor. Reasoning control and context score 0, with no reasoning and no stated context.

Granite Speech 5.0 2B 55

Chat template 20, chat_template.jinja in the repository (193 characters). Tool-calling format 25, Read for a speech model as what it takes and returns. Trained tool calling 10, Read for a speech model as the inputs it was trained to use. Structured output 0, Read for a speech model as typed outputs. Reasoning control 0, A speech model with no reasoning. Context 0, the language model is extended to 16K positions, per the card and config.json (16,384), below the 32K band, and the card says to send audio in segments of about two minutes. 20 + 25 + 10 + 0 + 0 + 0 = 55.