Published: 2026-09-23
Deep dive

Opus 5.5 vs GPT-6 Astra: 12 tasks, 17 hours of agent time, and a per-task cost sheet

Chapters / key moments (click to jump — plays here on the page)

This is the rare comparison video that reports time and cost on every single task, which is what makes it useful rather than merely entertaining. Both models were run on high effort in their own harnesses, on identical prompts. The quality result and the cost result point in opposite directions, and that tension is the actual finding.

Source video

"I Tested Opus 5.5 vs. GPT-6 Astra on 12 Real Use Cases" by Nate Herk — Watch on YouTube →

How the test was run

The same prompt was fired at each model inside its own harness, both on high effort. The work was done on subscription plans, but every task is also reported at API-equivalent billing, so the numbers mean something regardless of which plan you are on.

One methodology note worth repeating, because he raises it himself: in the previous day's Sol comparison, two experiments were spoiled when the two agents noticed each other and started coordinating. Here they were kept apart. If you run your own bake-offs, that is the trap to avoid.

The pricing setup

At list API rates, GPT-6 Astra is roughly 2.5× the per-token price of Opus 5.5. The framing question for the whole video is therefore not "which is better" but whether Astra delivers 2.5× the value. The measured answer is that it does not need to, because it burns far less wall-clock time and far fewer tokens reaching a comparable result.

Current published rates for both models are in our cost calculator, which is the number to plan against rather than any figure quoted in a video.

The totals

Measure	Claude Opus 5.5	GPT-6 Astra
Tasks won (his judgement)	8	4
Total agent run time	~11 hours	~6 hours
Total cost at API billing	$214	$132

Scoring is explicitly subjective — his taste, his use cases. The time and cost columns are not.

Where each one won

Opus took the creative and judgement work. Video editing (31 minutes / $10 against Astra's 39 minutes / $21), the Instagram reel, the branded deliverable set, the 3D learning world, and the Canva browser-use drawing test — where Astra's output was weak enough that he flagged it as a regression from earlier Astra runs.

Astra took the specified, bounded work. On the code review both models scored 100/100, but Astra got there in a fifth of the time at half the cost. It also won the trip itinerary ($11 against $14.47) and the Instagram carousel (about $7 against $10).

The pattern worth stealing

His summary: Claude models behave like a wise old owl — judgement and taste, and they cope with vague goals. GPT models behave like a very good obedient worker — give them an exact spec and they execute it efficiently, but vague goals frustrate them. That maps directly onto how you route work, not just which subscription you buy.

The budget argument he raises against himself

On the 3D-world task Opus ran 1 hour 44 minutes; Astra finished far sooner for about $12. His own question: if you gave Astra the same $60 of budget and let it iterate with feedback until it hit that ceiling, would it beat Opus's single pass? He thinks probably yes. A single-shot comparison systematically flatters the slower, more expensive model — worth remembering whenever you read a bake-off, including this one.

Key Takeaways

Run competing agents in separate working directories — agents that can see each other's work will start coordinating and spoil the test.
Route by task shape, not by leaderboard: vague and creative to Opus, tightly specified and bounded to Astra.
The 2.5× per-token price gap did not produce a 2.5× bill. Per-token price is a poor proxy for cost per task.
Both models cleared the bar on ordinary knowledge work; the differences showed up on taste and on long autonomous runs.
Single-pass comparisons favour the slower model. Budget-matched comparisons would likely read differently.

The companion test against the cheaper tier: Opus 5.5 vs GPT-6 Sol. Current rates for every model named here: cost calculator.

What you can actually set up from this

Extracted from the video's own transcript — the specifics the original summary left out.

Configuration shown

Results for the first three tasks shown in the video (speaker's numbers)

Landing page build:  Opus 5.5 40 min, $18.32  |  Astra 32 min, $11.33  -> Opus preferred on design
Sizzle reel from 105 GB of event recordings (HyperFrames): Opus 31 min, $10  |  Astra 39 min, about $22  -> Opus preferred
Instagram reel from a talking-head recording: Astra about half Opus's run time, about $9 vs $11  -> Opus preferred

Template shown in the video. Adapt to your project — do not copy verbatim without reviewing each line.

Reproducible steps

Same prompt, separate harnesses
Both models ran the identical prompt in their own harnesses (Claude Code for Opus 5.5, Codex for GPT-6 Astra), both on high effort and using /goal prompts. The speaker priced each run as if on API billing, though he used subscriptions.
Isolate each run in its own worktree
He admits that in the previous day's comparison two experiments let the agents work together, so this time each run gets a separate worktree so they cannot touch each other's files.
Record time and cost per task
For each of 12 tasks he logged how long each model ran and what it would have cost on the API, then judged the output himself.

Gotchas

Astra's token price is described as 2.5 times Opus 5.5's, yet it was cheaper on two of the first three tasks, so run time and number of turns matter as much as the price list.
Opus won the creative tasks (design, editing, reels) in the speaker's judgement; he says Astra felt bland and that its editing output seemed to have degraded since an earlier good run. That is one person's impression, not a benchmark.
Codex asked clarifying questions mid-/goal on the data-heavy task and kept the goal running; the speaker prefers that over Claude Code's behaviour.
The video covers 12 tasks but only the first few are summarised here; treat the totals as anecdotal with one run per model.