@cjzafir: 359M Tokens burned in 72 hours. Cost: $78~ Results: New 240M fine-tuning dataset. Process: > Codex 5.5 as Orchestrator.…
Summary
A developer used Codex 5.5 as an orchestrator and Deepseek v4 pro as an executor to generate a 240M token fine-tuning dataset, burning 359M tokens at a cost of only $78.
View Cached Full Text
Cached at: 05/15/26, 05:06 PM
359M Tokens burned in 72 hours.
Cost: $78~
Results: New 240M fine-tuning dataset.
Process: > Codex 5.5 as Orchestrator. >Deepseek v4 pro as Executor. (pipeline image attached)
Amazing times. https://t.co/nj5htyEqQU
Similar Articles
@Michaelzsguo: My Deepseek V4 Pro agent (inside codex) has been pursuing goal for more than 13 hours, burning ~100M tokens, and has on…
A user reports that their Deepseek V4 Pro agent running inside Codex has been operating for over 13 hours, consuming ~100M tokens but costing only $1.85, highlighting the model's cost efficiency.
@cjzafir: In the last 17 days, I burned $6k worth of tokens on my Claude 20x and Codex 20x subs. But I got so much work done by j…
The author open-sources a 'lean-thinking' skill for AI coding agents that reduces token usage by up to 35% while maintaining quality, based on personal testing with Claude and Codex models.
@cjzafir: Here's my Fine-tuning Dataset Generation Pipeline: > Codex 5.5 as an Orchestrator > Deepseek v4 Pro as a Generator In s…
The article describes a fine-tuning dataset generation pipeline using Codex 5.5 as orchestrator and Deepseek v4 Pro as generator, with autonomous quality gates and iterative improvement for high-quality synthetic data at low cost.
@VadimStrizheus: i just spent $260,000 in API credits on Codex in the last 24 hours I used up 5.2B tokens with GPT 6 Astra + GPT 5.6 Sol…
A user reports spending $260,000 in API credits on Codex within 24 hours, utilizing 5.2 billion tokens with GPT 6 Astra and GPT 5.6 Sol models.
My Codex run burned 100,000+ tokens over 8 hours and still handed me an unfinished mess... so I built something that just finishes
After a Codex run wasted over 100,000 tokens without completing a task, the author built a tool that reliably finishes code generation jobs.