Index-Translate-9B has 9.653 billion total parameters. The selected Q4_K_M GGUF file is 5.78 GB.
Data snapshot: . 3 published tests in this catalog.
Model file and details
Provider
IndexTeam
Total parameters
9,653,104,368
Exact file size
5,780,090,304 bytes
Quantization / precision
Q4_K_M
Weight repository
IndexTeam/Index-Translate-9B-GGUF
Original model repository
IndexTeam/Index-Translate-9B
Weight revision
645939fbab839a656f1bc39975a872bc8ad749f8
Model revision
4e06279e62352dab96bbeb6fa0579bca5e822d75
Runtime notes
Confirm model architecture and quantization support in your runner.
Recorded release date
2026-09-28
Date evidence
Official source repository created 28 September 2026; official GGUF package and expanded evaluation documentation reviewed after their 10 October 2026 revisions.
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.
FLORES COMET-22 · Index-Translate
87.89 / 100
Publisher-reported COMET-22 multiplied by 100 on 126,000 inputs covering 420 translation directions among 21 languages. Specialist translation quality test, not general intelligence or local inference speed.
Benchmark efficiency: 15.206 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-reported FLORES COMET-22 result of 0.8789 for the BF16 source checkpoint, shown as 87.89 on the catalog's 0-100 scale. The evaluation uses 126,000 inputs across 420 directions among 21 languages, greedy decoding and a 16,384-token context.
Source-checkpoint result, not an independent benchmark of the selected Q4_K_M. The official GGUF card reports A100 KL/RMS-probability checks and greedy spot checks against BF16, saying Q4_K_M outputs matched almost verbatim, but publishes no full Q4 score. Translation quality only; not speed or general capability. Apache-2.0 license. Multimodal image translation additionally needs a separate mmproj file, which is not included in the listed model bytes.
Publisher-reported reference-free GPT-5.6-Sol judge score averaged over 20 target languages, using zero-shot Chinese instructions, 3,584-token context and a 1,536-token output limit.
Benchmark efficiency: 13.036 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-reported WMT26 Judge result for the BF16 source checkpoint: zero-shot Chinese instructions, 3,584-token context, 1,536-token output limit and reference-free GPT-5.6-Sol scores averaged over 20 target languages.
Source-checkpoint publisher result, not an independent evaluation of the selected Q4_K_M. Reference-free judge scores depend on the evaluator and protocol. The official GGUF card provides qualitative parity checks rather than a full Q4 benchmark rerun.
Publisher-reported instruction-following score multiplied by 100 on 3,000 Chinese-to-20-language translation tasks across ten constraint types. GPT-5.6-Sol judges quality and adherence separately.
Benchmark efficiency: 14.202 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-reported instTrans IFscore of 0.8209 for the BF16 source checkpoint, shown as 82.09 on the catalog's 0-100 scale. The test contains 3,000 Chinese-to-20-language tasks across ten constraint types, with GPT-5.6-Sol judging translation quality and instruction adherence separately.
Source-checkpoint result, not an independent Q4_K_M test. Use the official constrained-translation prompt, greedy decoding and thinking disabled; ordinary chat prompts are not equivalent to the reported setup.
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 5.78 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 5.78 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 5.78 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.