# HELP atlas_generation_tokens_total Total tokens generated
# TYPE atlas_generation_tokens_total counter
atlas_generation_tokens_total 120
# HELP atlas_http_bytes_in_total Total HTTP request body bytes
# TYPE atlas_http_bytes_in_total counter
atlas_http_bytes_in_total 114
# HELP atlas_http_bytes_out_total Total HTTP response body bytes
# TYPE atlas_http_bytes_out_total counter
atlas_http_bytes_out_total 3772
# HELP atlas_prompt_tokens_total Total prompt tokens processed
# TYPE atlas_prompt_tokens_total counter
atlas_prompt_tokens_total 19
# HELP atlas_requests_active Currently active requests
# TYPE atlas_requests_active gauge
atlas_requests_active 0
# HELP atlas_requests_total Total requests processed
# TYPE atlas_requests_total counter
atlas_requests_total 1
# HELP atlas_spec_decode_verify_total MTP draft verify outcomes by K and result
# TYPE atlas_spec_decode_verify_total counter
atlas_spec_decode_verify_total{k="2",outcome="accept"} 5
# HELP atlas_time_to_first_token_seconds Time to first token
# TYPE atlas_time_to_first_token_seconds histogram
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="0.05"} 0
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="0.1"} 0
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="0.25"} 1
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="0.5"} 1
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="1"} 1
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="2.5"} 1
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="5"} 1
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="10"} 1
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="30"} 1
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="60"} 1
atlas_time_to_first_token_seconds_bucket{model="Qwen/Qwen3.6-35B-A3B-FP8",le="+Inf"} 1
atlas_time_to_first_token_seconds_sum{model="Qwen/Qwen3.6-35B-A3B-FP8"} 0.150161316
atlas_time_to_first_token_seconds_count{model="Qwen/Qwen3.6-35B-A3B-FP8"} 1
# HELP atlas_prefix_cache_hits_total Prefix cache lookups that found cached blocks
# TYPE atlas_prefix_cache_hits_total counter
atlas_prefix_cache_hits_total 0
# HELP atlas_prefix_cache_misses_total Prefix cache lookups with no match
# TYPE atlas_prefix_cache_misses_total counter
atlas_prefix_cache_misses_total 1
# HELP atlas_prefix_cache_hit_tokens_total Tokens reused from prefix cache
# TYPE atlas_prefix_cache_hit_tokens_total counter
atlas_prefix_cache_hit_tokens_total 0
# HELP atlas_prefix_cache_hit_rate Prefix cache hit rate (0-1)
# TYPE atlas_prefix_cache_hit_rate gauge
atlas_prefix_cache_hit_rate 0.0000
# HELP atlas_kernel_lookups_unresolved Kernel lookups that did not resolve for the live model
# TYPE atlas_kernel_lookups_unresolved gauge
atlas_kernel_lookups_unresolved 0
# HELP atlas_token_entropy_last Most recent per-token entropy (nats)
# TYPE atlas_token_entropy_last gauge
atlas_token_entropy_last 0.0000
# HELP atlas_low_entropy_tokens_total Tokens with entropy below 0.3
# TYPE atlas_low_entropy_tokens_total counter
atlas_low_entropy_tokens_total 103
# HELP atlas_low_entropy_ratio Fraction of tokens with entropy below 0.3
# TYPE atlas_low_entropy_ratio gauge
atlas_low_entropy_ratio 0.8512
