Best Fine Tuning Companies

ComparisonChecked Nov 8, 2026

Together AI vs Fireworks AI

Both run managed fine-tuning of open-weight models, publish per-token training prices, and let you download trained weights. They differ in how prices are structured, how reinforcement learning is offered, and how trained models are served.

What differs

  • Methods: Together AI lists LoRA and full fine-tuning, SFT and DPO pricing, and reinforcement learning; Fireworks AI lists LoRA and full-parameter SFT and DPO/ORPO, with RL through its Training API after removing managed RFT jobs.
  • Pricing: Together AI prices per 1M training tokens per model with a minimum charge per job ($4.00 to $60.00); Fireworks AI prices per 1M training tokens by model-size band (for example $0.50 LoRA SFT up to 16B) plus GPU-hour rates for dedicated training.
  • Weights: Together AI offers merged, adapter-only, or full weights downloads; Fireworks AI offers LoRA adapter download with firectl.
  • Serving: Together AI serves fine-tuned models on dedicated endpoints billed per minute per replica; Fireworks AI serves trained models on dedicated deployments with live merge or multi-LoRA.
  • Published security: Together AI lists SOC 2 Type II, ISO 27001:2022, and HIPAA-aligned options; Fireworks AI lists SOC 2 Type II, ISO 27001, 27701, 42001, and HIPAA with a BAA.

Who each tends to fit

Teams that want a per-model price list and full-weight downloads may start with Together AI’s pricing table. Teams that want to write custom training loops on the same platform may look at Fireworks AI’s Training API. Price the same dataset on both.

Last checked: . No winner declared.

Side by side

FactTogether AIFireworks AI
StatusActive, privately held.Active, privately held.
OwnershipPrivately heldPrivately held
Buying modelManaged fine-tuning APIManaged fine-tuning API
CategoriesManaged fine-tuning API; Inference deployment; Reinforcement fine-tuningManaged fine-tuning API; Training API; Inference deployment
PricingPer 1M training tokens (dataset tokens x epochs plus evaluation tokens), by model. Examples: Qwen3.5 0.8B SFT $0.34, DPO $0.84; Llama 3.3 70B SFT $2.03, DPO $5.08; Kimi K2.6 SFT $15.00, DPO $37.50. Minimum charge per job $4.00 to $60.00 depending on model.Managed training per 1M training tokens: up to 16B parameters $0.50 LoRA SFT, $1.00 LoRA DPO, $1.00 full SFT, $2.00 full DPO; 16.1B-80B $3.00 / $6.00 / $6.00 / $12.00; 80B-300B $6.00 / $12.00 / $12.00 / $24.00; over 300B $10.00 / $20.00 / $20.00 / $40.00. Dedicated Training API per GPU-hour: H100 and H200 $8.00, B200 $13.00, B300 $15.00, GB300 $20.00.
CertificationsSOC 2 Type II; ISO 27001:2022; HIPAA-aligned options (vendor wording; confirm BAA)SOC 2 Type II; ISO 27001; ISO 27701; ISO 42001; HIPAA (with BAA)
LimitationsPricing page has two tables and does not label which rates apply to LoRA vs full fine-tuning; confirm for your job; Dedicated endpoints bill per minute per running replica, even when idle; HIPAA wording is 'aligned'; confirm BAA availabilityTrained models serve on dedicated deployments; serverless per-token serving of your own adapter is not available; Managed RFT no longer accepts jobs; RL requires the Training API; Region-restricted dedicated training costs 1.5x
How to buytogether.aifireworks.ai
Sources[1] [2] [3] Checked 2026-11-08[1] [2] [3] [4] [5] Checked 2026-11-08

Facts come from each vendor’s profile sources. Run a paid pilot before choosing.