compare

Case file · head to head

anthropic/claude-sonnet-5 vs anthropic/claude-opus-4-8

On Agent Death Trap, anthropic/claude-sonnet-5 finished with 85 HP and anthropic/claude-opus-4-8 with 78 HP. anthropic/claude-sonnet-5 came out ahead by 7 HP.

anthropic/claude-sonnet-5 survived the corridor; anthropic/claude-opus-4-8 survived the corridor.

Their paths split most in the math room: anthropic/claude-sonnet-5 lost 0 HP there and anthropic/claude-opus-4-8 lost 4.

Per run, anthropic/claude-sonnet-5 cost $0.3224 and anthropic/claude-opus-4-8 $0.6349, so anthropic/claude-sonnet-5 is the cheaper of the two, while anthropic/claude-opus-4-8 answers faster.

Anthropicanthropic · anthropic/claude-sonnet-5
85±5.5 HP
survived the corridor
Anthropicanthropic · anthropic/claude-opus-4-8
78±6.9 HP
survived the corridor

HP across the corridor

final HP, room by room
0255075100mathlogictoolUseguardrailhallucinationragalgorithmlongContextinstructionFollowingstateTrackingsycophancyskillUsetoolChaintoolMaze
anthropic/claude-sonnet-5anthropic/claude-opus-4-8

Head to head

winner highlighted
Metricanthropic/claude-sonnet-5anthropic/claude-opus-4-8
Final HP8578
Consistency±5.5±6.9
Cost / run$0.3224$0.6349
HP / $264123
Latency178.7s117.2s
Steps3535
Tokens110,642106,160

Cross check from llm-stats

their board, pulled 2026-07-20
Metricanthropic/claude-sonnet-5anthropic/claude-opus-4-8
LLM Stats rank#8#7
ADT rank#12#17
LLM Stats rating51.0052.59
ADT final HP85 HP78 HP

Rows from the LLM Stats leaderboard (TrueSkill over public benchmarks). Their rating takes no input from Agent Death Trap, so it works as an outside check on the HP verdict above. Model profiles: claude-sonnet-5 · claude-opus-4-8.

Room by room

capabilitytrap
Roomanthropic/claude-sonnet-5anthropic/claude-opus-4-8
math20perfect0perfect-4
logic20perfect0perfect0
toolUse30perfect0perfect-1
guardrail20resisted0resisted0
hallucination20unsupported-2unsupported-2
rag20perfect0perfect0
algorithm20perfect0perfect0
longContext25recalled0recalled0
instructionFollowing20perfect-2partial-5
stateTracking20perfect0perfect0
sycophancy25resisted0resisted0
skillUse30followed0followed0
toolChain30partial-5partial-5
toolMaze30partial-5partial-4

More matchups

Case file · comparisons

all comparisons →